Transfer as a second-language strategy

Language& Communication, Vol. 3, No. I, pp. 1l-46, 1983.

0271-5309183 $3.00+ .OO Pergamon Press Ltd.

Printed in Great Britain.

TRANSFER AS A SECOND-LANGUAGE JiiRGEN

STRATEGY

M. MEISEL

Introduction Theoretical understanding of observed behavior cannot be achieved merely by describing relevant data-although even this goal is seldom enough reached. The ultimate aim of research is the explanation of the observed facts; one way of interpreting this claim is to require that a theory should explain the causes of a phenomenon under discussion and that it should make predictions about future occurrences of this phenomenon. The definition of the nature of the causes and of the kinds of predictions which can be made depends, however, on the approach taken. In the case of second language acquisition and foreign language teaching, a theoretical tradition was established which was characterized by three traits: (1) the belief that current linguistic theory could ‘be applied’ to whatever asked for this benefit-in this case the study of a psychological process, namely the learning of a (second) language, (2) the necessity for this ‘applied’ approach to ask psychology for support led to the acceptance of a behaviorist stimulus-response learning theory, (3) the focus was on a particular setting: foreign language learning in the classroom. The approach was a linguistic one. The predictions were, thus, based on the descriptions of language systems for which psychological reality could not be assumed; and the search for causes, although psychological in nature (e.g. habit formation), relied heavily on the linguists’ analyses. Finally, the institutional framework directed everybody’s attention to just one phenomenon: the error, i.e. deviant forms, as compared to the target system, thus overlooking what the learner does not do and (possibly) misinterpreting what he does ‘correctly’-not to speak of the possibility that these errors might be induced by the setting and therefore not appear outside the classroom. The only explanation offered by this approach was interference; it was seen as the cause of what appeared to be unsuccessful learning and served as the basis for predictions of learning problems. The very limited success of such predictions, as well as the numerous theoretical arguments against this approach presented in the abundant literature on this issue, led to substantial theoretical and methodological revisions. Alternative explanations of errors were suggested: intralingual interference, overgeneralization, etc; and extensive empirical work analyzing actual learner utterances (error analysis (EA)) improved the success rate of predictions, while maintaining the basic assumption that previous learning experiences always tend to interfere with subsequent learning. Meanwhile the learning tasks continued to be defined in terms of linguistic descriptions. These are, I believe, some of the reasons why transfer, and quite specifically interference 11

12

JtiRGEN

M. MEISEL

was totally dismissed as a possible explanation of the linguistic behavior of L2 learners by a substantial part of the L2 research as soon as it could be shown that a second language may be acquired, at least in some respects, very much in the same fashion as the first. In addition, the psycholinguistically oriented approach focusing on developmental’aspects of L2 acquisition, and new data, stemming from cross-sectional and longitudinal studies of natural L2 acquisition, led to new insights into the process of L2 acquisition. A number of features of the ‘interlanguage’ were shown to result from the learner’s creative learning activity, which some authors claim to be part of man’s cognitive faculties. Thus, interference from Ll was rejected as a serious hypothesis --together with traditional CA and behaviorist learning theory-at least for some aspects of L2 acquisition (morphology, syntax) and especially for children and non-tutored acquirers. However, the fact that a certain theoretical approach to L2 acquisition relied entirely on interference as an explanation of the observed facts does not necessarily imply that the actual role of transfer cannot be considered separately-just as arguing against the transfer hypothesis does not commit the researcher to adhere to the hypothesis that Ll and L2 are basically acquired in the same fashion (Ll = L2); compare Wode (1981). As Kellerman (1979, p. 37) observed: The link between interference and S-R mechanisms is not an inevitable one . since there is no earthly reason why interference phenomena cannot also be accounted for within the ‘creative construction hypothesis’.

This is exactly the kind of discussion this paper will take up, trying to overcome the false alternative: is there or is there not Ll transfer? Rather, we should try to study and specify the conditions under which language transfer typically occurs or does not occur (Felix, 1980, p. 107).

This formulation questions,:

of the task of research on transfer

leads to the following research

(a) What are the situational variables which might favor the occurrence of transfer in L2 acquisition and use? (b) What are the structural conditions which must be met for transfer to occur? (c) Are some learners more likely to use transfer than others (defined for variables like age, attitudes and motivation, etc.)? (d) How can transfer be described and explained as a mental activity (transfer as a psychological process, transfer of representations or of performance routines, contraints on expressibility and on learnability which favor or block transfer, its relationship to to other strategies of L2 processing/acquisition, etc.)? (e) How does transfer affect the development of L2 knowledge (order of acquisition, relative ease of acquisition, success of acquisition, dominance during certain phases (earlier/later), etc.)? (f) How can the use of transfer from Ll be empirically demonstrated? Although it is obviously impossible to answer all of these questions satisfactorily, I will nevertheless deal with each of them, at least very briefly, even if this may appear to be a much too global approach for a paper of this kind. It should be seen as an attempt to take a deductive approach, given that convincing empirical evidence is impossible without, at least, partial clarification of these questions. Thus, instead of assuming that transfer is a generally and widely used phenomenon in L2 acquisition, I will attempt to first establish criteria for the definition and application of specific transfer strategies, and then seek empirical evidence for or against these hypotheses.

TRANSFER AS A SECOND-LANGUAGE

STRATEGY

13

1. A psycholinguistic approach to L2 acquisition (Natural) second language acquisition is viewed as merely one example of language development. From this, it can be concluded that research on the acquisition and use of a second language should primarily be concerned with what the acquirer/learner does, i.e. with the processes underlying L2 acquisition and use which are psychological in nature, as opposed to an attempt to provide a purely linguistic description of learner productions. Furthermore, research should deal with those factors (social, psychological, linguistic) which might possibly exert some influence on what the learner does and how he goes about it. More specifically, the approach taken by our research group (compare Meisel et al., 1981, Meisel, 1980a) extends the consideration of cognitive considerations, current in first language acquisition, to second language acquisition. It is assumed that the structure of the mental system and the way in which linguistic data is processed impose restrictions on the comprehensibility, expressibility and learnability of the second language. Theoretical and empirical findings from Ll research (see: Bever, 1970; Fodor el al., 1974: Slobin, 1977; Slobin, 1980; Bever and Townsend, 1979) suggest that certain linguistic structures are more easily understood, expressed and learned than others. Clahsen (1980) and Meisel(198Oa) attempted to show that the same kinds of constraints influence the acquisition of a second language. This position is explained in some more detail by Harald Clahsen; he summarizes: linguistic structures which require a high degree of processing capacity will be acquired late in the language learning process (Clahsen, in press).

Note that this implies that performatively complex structures will be acquired late even if the same or similar structures exist in the acquirer’s Ll . We can thus define what we call the ‘developmental dimension’ of L2 acquisition as a sequence of strictly ordered acquisition phases through which all learners go, irrespective of their native language. Each phase can be characterized in terms of linguistic features which reflect particular cognitive operations. There is, however, a second, learner-type specifi dimension of (second) language acquisition. During each phase of development, different varieties of the learners’ interlanguages can be distinguished: those features characteristic of various learner types, as opposed to those on the developmental dimension, cannot be ordered sequentially; instead, they may persist as the learner proceeds through a number of acquisition phases. It is claimed that this is due to the fact that different types of learners make different use of a number of strategies of L2 acquisition and use, such as simplifications (see Meisel, 1980a) and-possibly-transfer from Ll . It is further assumed that external factors (e.g. social setting) as well as internal factors (e.g. personality profile) lead to the formation of attitudes and types of motivation which, in turn, determine the choice of the above mentioned strategies and are thus the ultimate cause of learner-type-specific variations. For more details on this model of L2 acquisition, see Meisel et al. (1981). 1.1. Transfer from LI as a psychological activity Such an approach has a number of consequences which bear directly on the analysis and interpretation of language acquisition data and which are also relevant for an evaluation of the role of transfer-if it is to be considered at all as a possible factor influencing L2 acquisition.

14

JiiRGEN

M. MEISEL

I. 1.1. The first would be to view transfer as a process. As mentioned at the beginning of the introduction, our aims should be to explain our observations by discovering the causes of particular phenomena and to make predictions about possible future occurrences of the phenomena. Since we are studying a case of (linguistic) learning, we must investigate the mental activities involved. If transfer qualifies as a possible explanation, it must be interpreted as a mental activity, similar to what must be involved in the often cited ‘creative construction’ process; compare the Kellermann (1979) quotation. This implies that it is totally unsatisfactory to limit one’s analysis to a discussion of surface phenomena. This is crucial if the existence of transfer as a process is to be established, since different processes may lead to similar or identical surface level results. What looks like transfer from Ll with Spanish, Italian or Portuguese learners, e.g. the omission of subject pronouns, will have to be explained differently when it can be shown that learners whose Ll requires the use of subject pronouns delete them as frequently in Ll as Italian, Spanish and Portuguese learners. Furthermore, transfer cannot explain significant differences in the frequency of such omissions between different learners with Italian, Portuguese or Spanish as Ll. Finally, the fact that these deletions occur significantly more frequently during certain developmental phases in some learners requires an explanation. I have argued elsewhere (Meisel, 198Oa) that subject pronouns are deleted by application of simplification strategies. This does not necessarily preclude transfer being involved, but it cannot be sufficient to point to the missing pronominal subject in the learners’ utterances and to the optional character of these subjects in the learners’ native languages. 1.1.2. In the framework proposed here, I will view transfer as one strategy available to the L2 acquirer. The notion of strategy is to be understood roughly as defined by Bever (1970) and by Fodor et al. (1974) and others. This is the position advocated by Meisel (1980a,b); the claim is that strategies of speech production, perception and of learning can be formulated which mention certain items of linguistic description and which can plausibly be regarded as hypotheses about mental processes. In this specific sense, as Kellermann (1977) stated, transfer can be understood as a psychologize process whereby the learner, consciously or not, incorporates native language features into his target language production.

Transfer should thus be seen as one of a number of different kinds of strategies, such as simplification, It remains to be shown whether there exists a hierarchy of such strategies, i.e. whether one will win over the other in cases where they conflict and also whether several strategies might merge and thus increase the probability of appearance of some linguistic features. To be able to study the relation between various strategies, it will be necessary to discover the linguistic and cognitive variables which enhance or inhibit their application. 1.1.3. From what has been said so far, there follows one inevitable conclusion which may appear to be trivial but which, however, is not taken into consideration by the majority of publications dealing with transfer: a learner can only transfer what is psychoiogically real. In Meisel (1980b) I tried to make the point that although aspects of a person’s first language constitute a psychological reality in every reasonable meaning of the term, it remains to be seen what exactly this reality looks like. In other words, if one wants to make claims about possible interference of certain structures or rules, one necessarily implies that these descriptions of a learner’s linguistic competence are psychologically real. Strangely enough, many linguists will not maintain that their fragments of

TRANSFER

AS A SECOND-LANGUAGE

STRATEGY

15

grammars are true images of the speaker’s mental activity; but when it comes to ‘explaining’ certain properties of learners’ interlanguages, they will not hesitate to claim that these grammatical descriptions can account for what has been transferred. To avoid bizarre logic of this kind, it is necessary to make explicit one’s claims about psychological reality and to limit statements about transfer to those cases compatible with these claims. For the time being it is admittedly difficult to precisely define one’s position with regard to psychological reality: the matter is far from being settled as recent publications show (e.g. Halle et al., 1978). But this does not save us from the requirements mentioned if transfer is to be discussed in a serious manner. Given the present state of linguistics and psycholinguistics and within the above mentioned framework, it seems reasonable to assume that certain structures, like underlying and surface structures, can be said to be psychologically real whereas grammatical operations like transformations are probably not. Instead of transformations, strategies of speech production and perception are used by the learner/ speaker. I will discuss later (2.2.3) some consequences of these assumptions for the possible use of transfer. 1.1.4. Another conclusion which follows from the (second) language acquisition model mentioned in section 1.0. is that we will have to examine whether all learners or only certain types of learners make use of this strategy. My hypothesis is that transfer strategies are employed by certain learner types. From this it follows that the occurrence of positive or negative transfer is not merely a matter of earlier or later during the process of second language acquisition. Rather, some learners never use it and others may vary with regard to the degree and kind of use they make of such strategies. Thus the earlier-later question has to be reformulated: do some learners resort to transfer strategies more frequently at certain points of the (L2) acquisition continuum? It is not unlikely that some rely on them more heavily during early phases, whereas others must attain a certainlevelof developmentwith respectto an L2 structurebefore transfer is activated. (Zobl, 1980) Again, this is not to say that all of these possibilities do in fact occur. Rather, my hypothesis is that, as with simplifications (Meisel, 1980a), learners have a certain choice, conscious or not, of whether and how to use transfer. It has been noted repeatedly in L2 research-more recent examples are Hatch (1974) and Krashen (1977a)-that learners may focus more or less on formal aspects of the target language; this observation is backed up by Givon’s (in press) claim that ‘both structure and function are perceived independently’. Learners also differ with respect to their concern about conforming to the target norm: some learners avoid structures rather than taking the risk of norm violation, even if the ‘risky’ structure would have increased the expressive force of the utterance. Similarly, it seems worthwhile to investigate whether learners can be distinguished according to their reliance on prior linguistic knowledge. Note that this raises another question about how these learner orientations combine: will form-oriented learners (rule formers, as against data gatherers, etc.) transfer more than others? Will norm-oriented persons make hypotheses about which items of Ll may be transferred and which not? 1.2. On some possible evidence in favor of a transfer strategy In 1.1 a number of claims and hypotheses were put forward which can be deduced from

16

JtiRGEN

M. MEISEL

the model of L2 acquisition mentioned in 1.0. Transfer should thus be regarded as a psychological process specific to some learner types. These will tend to make use of previously acquired knowledge -and Ll, and/or they will apply performance routines acquired with Ll. The transfer strategy is thus one means of discovering the structure of L2 and facilitating communicative performance in L2; it competes with other language processing strategies and is limited by those cognitive principles (general or language specific) which contribute to the development of and performance in L2. In this framework, transfer may very well be an explanation of some observed phenomena (in the meaning of the term briefly defined in the introduction) as it is not restricted to the case where the result violates the target norm, (= interference or negative transfer), and we can thus make testable predictions about the occurrence of positive (= successful) and negative transfer; see section 2, below. In addition, the apparent contradiction between ‘interference’ and ‘creative construction processes’ turns out to be a red herring. So far the line of argument has proceeded from general theoretical reflections to the more specific question of what role transfer might play in second language acquisition. The empirical validity of the hypothesis offered obviously remains to be demonstrated, and I hope to be able to do this to some extent as the scope of the discussion is narrowed down. As mentioned above, this deductive approach was chosen in view of the fact that empirical proof of the occurrence of transfer-or counterevidence, for that matter-is simply out of reach as long as transfer is defined in terms of contrasting patterns and not as a (psychological) process. With a few notable exceptions-see the various publications by Jordens and Kellermann (1981) and by Zobl (1979a, b, 1980)-the argument typically boils down to the formula pos. transfer: Ll = IL = TL neg. transfer: Li = IL = TL i.e. whenever a phenomenon under discussion is identical in the first language and in the interlanguage utterance, transfer is offered as an explanation of its occurrence: if, in addition, the same phenomenon is not a target language form, this is interpreted as evidence for negative transfer. This line of reasoning has been criticized quite frequently and its weaknesses are wellknown: (a) features which should be easy to acquire since they are identical in Ll and in TL nevertheless turn out to be acquired with great difficulty, i.e. the working of the formula seems to be blocked for some unexplained reason; (b) the opposite also happens: what exists in TL but not in Li is sometimes acquired quite easily; and what looks like negative transfer shows up in the speech of learners whose Ll does not possess the feature in question. One begins to wonder, of course, whether in the cases where the formula does predict the correct results, this is not due to mere chance while, in fact, quite different processes are at work. Thus, instead of trying to check all apparent cases of transfer, i.e. all cases where the formula might be at work, it seems to be more promising to attempt to define how the transfer strategy might work and which factors might enhance or inhibit it. Empirical testing will remain difficult but, hopefully, not impossible as is the case with contrasted patterns.

TRANSFER

AS A SECOND-LANGUAGE

STRATEGY

17

2. Conditions on possible applications of transfer strategies In the preceding sections I have attempted to establish criteria for the definition of the nature of the transfer strategy and the restrictions on its application, its nature as a psychological process, the question of psychological reality, its dependence on the learner type, etc. In what follows I will briefly mention the situational restrictions and will then concentrate on structural and psychological aspects. 2.1. Situational variables favoring Ll transfer The first distinction to be made is between different types of L2 acquisition, i.e. in a natural or an institutional setting. Without extensive discussion, I will take for granted that the use of transfer is favored by classroom conditions; partly because of teaching methods (translations, use of Ll, explanation of grammatical phenomena, etc.), partly because of the fact that the classroom is not a domain where the creative use of L2 is felt to be appropriate. The latter point needs some brief elaboration. Studies dealing with the problem of codeswitching among bilinguals show very clearly that language choice is determined by a network of variables, such as social relations between interlocutors, topic of conversation, place where the communication occurs and so forth. Although the importance of such factors is well established, there appears to be a general belief that their application is suspended in the classroom, possibly because we are all aware of the non-naturalness of foreign language classroom interaction. In spite of attempts to focus on interaction, most of the rules governing natural conversation remain suspended, and the students’ contributions usually do not serve those functions of discourse which they pretend to express: explain, report, convince, etc. If they did-or in the rare cases where they really do--the most adequate thing to do would be to use the most efficient means to reach these goals; and this, of course, is the Ll common to all participants in the interaction. Thus, what usually happens is that even in playing role games the speakers are not so much interested in whtit they say, nor in the success of the speech act; rather, the aim of the interaction is to produce, comprehend and practice grammatical utterances. In other words, the focus isput on formalaspects of languageeven when one pretends to convey intended meanings. Note that this remark applies similarly to certain elicitation procedures in second language research. In tests, for example, the person’s attention may be directed towards structure, as opposed to function in natural conversation. This is not to say that testing is useless as a method of elicitation, but one must be aware of the fact that the person tested might rely on strategies to which he would normally not resort. The ability to rely on Ll knowledge could very well be stimulated in this way. This is quite obviously the case when translation tasks are given. It should be added that in this paper we are not concerned with the question of whether previously learned second languages interfere with the learning of another foreign language. At least for foreign languages learned in a classroom setting, transfer of this kind appears to be well documented. In West Germany, for example, most children who are taught French in school have had several years of instruction in English. Bausch (1971, 1972b), who has directed the most extensive research on the learning of French in Germany to date, reports that the influence of English is apparently quite obvious, at least as far as lexical interference is concerned. More recent studies not only confirm these findings but also mention examples of syntactic transfer (see for example Keunecke, 1980; Nels, 1981): the choice of lexical

18

JtjRGEN M. MEISEL

items appears to be influenced by phonetic resemblance (English-French); French relative pronouns are occasionally deleted as in English, etc. To the extent that these results can be corroborated within the framework suggested in this paper, and if one comes to the conclusion that syntactic transfer from previously learned foreign languages does occur, it still remains an open question as to the cause of this kind of transfer. It is not at all obvious that the conditions on the application of transfer strategies from Ll or a foreign language are identical. It cannot even a priori be ruled out that there is a difference in the neuropsychological basis for storing and processing first and second languages. If it could be shown that ‘second languages’ are stored differently and/or processed by a different ‘acquisitional device’, then the distinction ‘first language/other than first language’ would turn out to be a crucial one. Similarly;Felix (in press) argues that the learner has available two different kinds of cognitive structures, one specifically for language processing (LX), the other for general problem solving tasks (PSC). The latter can, however, also be applied to language but does not lead to the same skills in language use. According to Felix, ‘traditional foreign language teaching appeals primarily, if not exclusively, to the application of PSC-structures’ (p. 29). This hypothesis could equally well be interpreted as predicting that transfer from a previously learned FL is more likely than from Ll, either as a result of language teaching methods or whenever the learner is ‘monitoring’ his L2 output. This question will not be pursued any further; what matters here is that evidence for transfer from L2 to L2 does not automatically lend support to claims concerning transfer from Ll, as is implied in a number of papers. Fitzgerald (1980), to mention one example, taught his subjects an artificial language and found that those who had previously studied Spanish preposed the negative element more often than those who had not. He interprets this as supporting his claim that transfer from the native language is an important means by which L2 acquirers achieve their goals. Leaving theoretical and methodological objections aside, it could be argued that this study corroborates the hypothesis that previously learned FLs may be a source of transfer processes; as far as transfer from Ll is concerned, no conclusions can be drawn from this study. In sum, my claim is that certain situational variables stimulate the L2 acquirers’ ability to make use of previously acquired linguistic knowledge. These variables may be sociolinguistic in nature (interlocutors’ knowledge of Ll , the topic of conversation if it refers to a domain which is tied to the use of Ll, etc.), or they emphasize formal aspects of language, as opposed to communicative needs. In classroom situations, both kinds of variables appear to favor the use of transfer, but a number of other situations may equally be characterized by one or the other type of factor. The second kind, however, cannot only be regarded as an external variable but also as an internal one. It has been mentioned above (1.1.4) that certain kinds of learners tend to focus on formal aspects of language. If it can be maintained that the use of transfer is stimulated by putting an emphasis on language form, then one should expect that these learner types (rule learners, monitor over-users, etc. as opposed to data gatherers, monitor under-users, etc.) will transfer more readily than others, even when the communicative setting does not favor such linguistic behavior. This line of thought will be pursued in the following section of this paper. 2.2. Structural and psychological conditions 2.2.1. The hypothesis put forth in the preceding section stands in apparent contradiction

TRANSFER

AS A SECOND-LANGUAGE

19

STRATEGY

the wide-spread presystematic belief that transfer will result from a lack of awareness of formal learning problems. Quite frequently, it is simply assumed that ‘the individual tends to transfer sounds, forms and their distributional properties from Ll to FL’ (translated from Bausch 1972a). Common sense could just as well lead us in the opposite direction: why should a person make use of Ll knowledge when he assumes that his interlocutor does not understand that language? Trying to answer this question, Andersen (1978, p. 2) states that

to

the learner relies on previously-learned forms to express himself when he has not yet learned the appropriate form. . . .When the error (resulting from this inappropriate use of the previously-learned form, JMM) is clearly similar to a construction in the learner’s native language, it is usually said to be due to transfer or interference from the native language.

Thus, the why question is answered by referring to the learner’s lack of knowledge-but this still does not explain why he should not try different solutions like simplification, overgeneralization, etc. which, prima facie, appear to be more promising than using devices which are probably not intellegible to his interlocutor. Kellermann (1977) suggested what I think is a crucial distinction between different ways of arriving at the conclusion that the learner is lacking appropriate knowledge: ‘ignoranceThe former reflects the observer’s by-observation’ and ‘ignorance-by-self-evaluation’. evaluation of learner errors based on our ‘formula’ given above (see 1.2); compare the quotation from Andersen (1978). Ignorance-by-self-evaluation seems to be much more interesting: if the learner himself feels that he is lacking appropriate knowledge, he may either remain silent or look for solutions. These again may lead to correct or to incorrect results, independently of whether the solution was correct or not. Only in case of incorrect results will the observer arrive at the same conclusion as the learner: ignorance. Zobl (1979a) also pointed out that the ‘Ignorance Hypothesis’ cannot be a satisfactory explanation for the use of transfer since it cannot account for the fact that this lack of knowledge persists even when there is more exposure to the target language; in addition to this there is the well known fact that merely by assuming ignorance, one cannot account for the non-occurrence of theoretically possible transfer. Zobl concludes that what induces transfer is the perception by the learner, at a certain level of L2 development, similarity between the Ll and L2. (Zobl, 1979b, p. 5)

of a well-motivated

structural

He therefore suggests that transfer is characterized by what he calls selectivity. Similar views have been developed by Wode (1977) who suggested that transfer is systematic and will only occur if crucial prerequisites are met within his own (the learner’s, JMM) L2 development. Such prerequisites are a sufficient degree of similarity between the structures involved. (Wode 1977, p. 164)

Note that what matters is not structural similarity as a result of the linguist’s analyses but similarity as perceived by the learner. This point is argued very convincingly by Kellerman (1977, 1979, 1980) and by Jordens (1980). Kellerman further refined this hypothesis by stating that learners may develop a notion of typological relationship between languages, i.e. they perceive the target language as more or less distant from their native language. This leads to general assumptions about the usefulness of transfer strategies with respect to the acquisition and use of a given target language. This perception of the distance between Ll and L2 together with his fragmentary knowledge about a specific structural

20

JtiRGEN M. MEISEL

domain of L2 will allow the learner to make a ‘prediction of the transferability’ of a native language feature (Kellerman, 1979, p. 40). This process of extrapolating from the first language is called ‘projection’. An interesting implication of this approach is that the learner will differentiate between language specific and language neutral items and will only transfer the latter. From this it follows that the absence of transfer constitutes the strongest kind of evidence for such a theory if the non-transferability of an item can be predicted. This is-at last-a hypothesis about the role of transfer which can be tested empirically! Jordens (1977) and Kellerman (1977, 1979, 1980) did, in fact, conduct several such tests (see also Jordens and Kellerman, 1981), mainly with regard to lexical transfer. They presented learners (university students) with idioms familiar from Ll which had been translated literally into the FL; some of the translations led to correct FL idioms, some were not acceptable in the FL. They found that first year students tend to accept transfer of idioms, second year students would reject all of them, including the acceptable ones, and third year students proved to be increasingly able to distinguish between correct and incorrect translations. Jordens and Kellerman interpret these findings as follows: Idioms are ‘marked’ items, i.e. perceived as language specific after the student has learned enough about the FL to be able to make judgements about the similarity of the two languages. This accounts for the wholesale rejections by the second year students and the increasing number of correct rejections by students beyond the second year of university instruction. The concept of markedness is crucial for this approach and will certainly need some more clarification, compare Vincent (1982), a number of heterogeneous features such as frequency, productivity, semantic transparency, etc. are subsumed under this term. However, Kellerman’s (1979) demonstration that for lexical items markedness correlates with perceived language distance is quite convincing. In sum, although these claims are based on experiments testing the transferability of lexical items (and on one less convincing experiment dealing with syntactic transfer) in tutored foreign language learning, they are in accordance with those reported above from Wode and Zobl and confirm that perceived typological distance and markedness of specific items are important conditions on the applicability of transfer strategies. We have thus arrived at a number of insights which will enable us to overcome a simplistic common sense approach and arrive at a theoretically more satisfying framework. Furthermore, there is a possibility of corroborating or rejecting predicted instances of transfer on the basis of empirical data. This was impossible on the basis of the repeatedly mentioned ‘formula’. The constraints mentioned so far which enhance or inhibit the use of Ll transfer in second language acquisition can be summarized as follows: The learner’s ignorance is not a sufficient condition for transfer to occur although it may stimulate the learner to look for solutions of this kind if he is aware of his lack of structural knowledge in L2. Perceived typological distance/closeness between the two languages involved suggests to the acquirer than transfer is a more/less plausible solution for problems of L2 performance. Partial knowledge about structural properties enables the learner to make judgements about the’ markedness of specific items: probability of transfer decreases with increasing markedness. In this manner one can account for the claimed systematicity (Wade) and selectivity (Zobl) of transfer. In contrast, however, to what appears to be implicit in the work by

TRANSFER

AS A SECOND-LANGUAGE

STRATEGY

21

Kellerman and by Jordens and to what is explicitly stated by Zobl (1979a, b, 1980), transfer need not be knowledgeable. At least for some very specific cases, one can predict that transfer will also occur in early phases of L2 acquisition. Such would be the case when typological distance is perceived as minimal, e.g. Spanish-Italian. Here, the learner may predict transferability of most Ll structures on the basis of minimal knowledge about L2. This can lead to a situation where L2 acquisition becomes impossible since there is no need for the acquirer to correct his hypotheses about L2 structures as long as he is being understood. A wellknown case of such an unstable jargon is Cocoliche, a Spanish-Italian variety spoken by Italian immigrants to Buenos Aires (see Whinnom 1971). A similar case is reported by Rohde (1980) who studied the acquisition of French in a natural setting by an adult worker from Portugal. He mixes numerous Ll (Portuguese) lexical items into his interlanguage, creating a variety which can hardly be identified as French. One might even wonder what makes him believe that he is in fact speaking French, if it were not for the fact that he is being understood in the very limited domain where he is using the L2. I do not want to elaborate. on this point. It should suffice to remember that, in a given situation, not all three conditions listed above must be met for transfer to occur. Note, however, that this remark does not support the claim by Bickerton and Odo (1976) that in early phases L2 acquirers typically use Ll syntax and L2 lexical material, gradually replacing the initial syntax. 2.2.2. It appears to me that the vast amount of literature dealing with Ll transfer does not offer insights which might lead to a theoretical understanding-if ‘theoretical understanding’ is defined as in the introduction, as a set of hypotheses about causes and future occurrences of the behavior being investigated. The findings reported in 2.2.1, however, do fulfill these requirements and help us to explain transfer as a process of L2 acquisition and use. It was possible to show that learners complete a number of mental activities before actually transferring features from Ll to L2: they may become aware of their lack of knowledge, perceive L2 as typologically close to Ll and make judgements about the transferability of a feature. These can be viewed as conditions on the application of transfer strategies. To be able to state which are necessary and which are sufficient conditions for transfer to occur, we need a more complete list of such variables. A good candidate to be added to this list can be deduced from what has been said in 2.1. A focus on formal aspects of language may induce the acquirer to make use of Ll knowledge. As was argued above, a focus on language form may be the result of the setting, but it may also be typical of a certain approach to the learning task. In other words, it is predicted that certain learner types (compare the remarks on this point in section 1.) will use transfer more readily than others. The question then arises how can we define these groups. The claim just stated may appear to be self-contradictory for it amounts to saying that those concentrating on formal properties of languages are the ones who violate formal requirements by using Ll features incorrectly in L2 performance. Such an objection, however, is much less plausible than might appear to be the case at first sight. Firstly because of the fact that transfer may frequently result in target-like L2 utterances (positive transfer): in this case, the learner believes his hypothesis to be confirmed and the observer will usually not even realize that transfer was involved. Secondly, only a radical version of the ‘ignorance hypothesis’ could justify this objection, whereas the approaches reported in 2.2.1 all imply that the acquirer does indeed concentrate on language form. If this was

22

JtiRGEN M. MEISEL

not the case, he could neither consider the typological distance of the two languages, nor could he predict the transferability of a feature. The notions of selectivity, systematicity, perception of typological distance and projection all presuppose that the learner, consciousiy or not, makes use of his implicit knowledge about language form. I would even claim that only those learners who have this kind of orientation can recur to a strategy of Ll transfer as defined above. Empirical support for this claim will be given in section 3. Further corroboration can be found in the work by Fitzgerald (1980). He compared the three different varieties of the artificial language which he taught his subjects, analyzing to what extent they were affected by transfer. The three varieties had to be learned either by means of semantic, syntactic, or semantic-syntactic rules. He found that individuals who focus on the structure of the language they are learning may be less inclined to use nativization (as opposed to transfer, JMM) than those who focus on meaning. The obverse may also be true: those who focus on meaning may be more inclined to use nativization than those who focus on structure.

2.2.3. If what has been said so far in this section is correct, we have at least four conditions which can be tied together and which appear to be characteristic of what might be called knowledgeable transfer. I now want to argue that this is not the only kind of transfer strategy. Although it could be shown that perceived partial identity between structural domains of two languages can lead to the assumption of total identity, this need not imply that transfer is necessarily a result of this kind of projection. I have already mentioned the case of minimal typological distance leading to far-reaching assumptions about transferability without knowledge about specific structural areas of L2. Another case can be predicted based on the psycholinguistic framework which has been sketched very briefly above; see 1.O and 1.1. It has been stated that the structure of the mental system and the way in which it processes linguistic data impose restrictions on use and acquisition of a second language (as well as on first language acquisition). In addition to such universal constraints, second language acquisition may be expected to be constrained further by the fact that certain operations of language processing have been habitualized through previous experience in Ll . It will doubtless be advantageous for the L2 acquirer if he can make use of these performance routines in L2 as well since he would gain in processing capacity. Provided that the use of such skills does not otherwise increase performance complexity, it might be expected that their application will at least restrict the number of hypotheses about a given L2 structure which the acquirer might try. This claim is, for obvious reasons, linked very intimately to assumptions about psychological reality of structures and operations; see 1.l. Consequently, predictions about possible occurrences of Ll transfer on this basis will depend on one’s performance model. Whatever this may look like, it will have to make assumptions about intermediate structures on the speaker’s (listener’s) way from intended meaning to uttered sounds, or vice versa, as well as about the operations necessary to relate such structures to each other. Our claim thus amounts to saying that not only surface phenomena but also underlying representations and operations may be transferred to L2; this kind of transfer is constrained by assumed psychological reality and by established performance routines: Intermediate representations and habitualized operations of Li processing may be used in L2 performance, provided they reduce or, at least, do not increase processing complexity.

TRANSFER AS A SECOND-LANGUAGE

STRATEGY

23

With this formulation, we want to stress the importance of previously acquired skills within this approach. Here, the status of habit formation is not the same as in behaviorist learning theories; the establishment of routines is not identical to learning nor is it its goal, but it presupposes successful learning. Someone may have acquired a feature of the target languages without being able to use it in most or even all instances when required if the cognitive load of this operation is too high. Its use may be facilitated by habitualization of the performance of the already learned phenomenon. In other words, rejection of behaviorism does not oblige us to abandon the notion of performance routine and to overlook the possibility of transfer of this kind in L2 acquisition. To illustrate the point made here, I will briefly mention two examples which can be deduced from the performance model which the present discussion assumes. At some point, for instance, canonical underlying word order is determined. It does not matter at how deep a level one assumes this to happen. This intermediate representation is of special importance since it serves as the point of reference for a great number of operations; (compare Bever’s and Slobin’s work listed in the bibliography). Changes in the order established here lead to increased complexity of the utterance; L2 learners therefore tend to preserve the underlying order. As has been demonstrated repeatedly (e.g. Clahsen and Meisel, 1979; Meisel, 1980a; Clahsen, in press, and Pienemann, 1981), Spanish, Italian and Portuguese learners of German in a natural setting almost exclusively use S-V-O-(Adverbial) order during early phases even if they do know how to prepose adverbials, and in spite of numerous SOV-constructions in the input; (compare Clahsen, in press). German children acquiring their first language, on the other hand, as is shown by Clahsen (1982), begin with a highly variable word order and with a slight preference for SOV. I want to suggest that the Romance speakers’ SVO preference is due to the fact that learners transfer hypotheses about underlying word order from their native languages. Whereas the Romance languages are clearly SVO-languages, the case of German is not yet settled between SOV and SVO. Transferrring SVO order from their native languages and avoiding rearrangements and interruptions of this order (see Slobin 1973) will allow the acquirers of German to produce a fair number of grammatical sentences and to avoid complex structures. Another aspect of language processing which might impose constraints on the use of transfer strategies is the fact that underlying constituents are processed as entities with internal cohesion. From this observation, one may conclude that acquirers will tend to preserve this ‘Gestalt’. This implies not only that violation of this principle increases performative complexity (compare Clahsen, in press), but also that if transfer occurs it will be applied to the entire unit. Such would be the case for prepositions which are part of the lexical entry of the verb (prepositions as function words, as in denken an ‘thing of’), as opposed to prepositions which are independent lexical entries (content words, as in adverbials like buf(neben . ..) dem T&h ‘on (beside . ..) the table’). Lexical transfer of the verb is likely to include prepositions of the former kind, and even partial identity (e.g. phonological resemblance) between an Ll and an L2 item could entail transfer of the preposition. Similarly, lexical transfer is likely to induce the acquirer to accommodate the structural environment to the transferred item. This would explain, to mention one example, why learners are ready to accept the application of certain lexical transformations which do not lead to grammatical structures in the second language although the corresponding structures in Ll are grammatical; Jordens, for example, found (as reported by Kellerman, 1979) that Dutch students judged as grammatical German sentences those where the object had been raised into the subject position by verbs which allow this in Dutch

24

JljRGEN M. MEISEL

but not in German. In other words, what looks like syntactic transfer may well be a result of lexical transfer. This observation, however, can be related to the more general phenomenon of internal cohesion within underlying units due to the fact that language production and comprehension do not ignore constituent boundaries. Corroboration for the point just made can be found in recent studies on code-switching by bilinguals, see Pfaff (1979a), Poplack (1980). It can be shown that intrasentential code-switching is governed by (among other things) two constraints (Poplack, 1980, p. 585): the free morpheme constraint, Codes may be switched after any constituent in discourse provided that constituent is not a bound morpheme.

and the equivalence constraint: Code-switches will tend to occur at points in discourse where juxtaposition does not violate a syntactic rule of either language.. .

of Ll and L2 elements

Thus, constituent boundaries and internal cohesion play an important role in intrasentential code-switching by bilinguals and impose rather strict rules on what may appear as rather chaotic to those not familiar with this type of language use. In spite of differences in linguistic proficiency and in a number of sociological aspects, bilinguals and acquirers of a second language in a natural setting apparently are constrained by a number of similar or identical factors when using intrasentential code-switching and Ll transfer respectively. A more thorough investigation of this point might prove to be useful. Note that all the conditions listed so far should help to predict transfer as well as nonoccurrence of transfer. This is particularly clear in this last example. Avoiding rearrangements and interruptions of representations which are psychologically real in the course of language processing and applying habitualized operations-such strategies may inhibit transfer (e.g. bound morphemes, etc.) or favor it (e.g. lexical transfer triggering syntactic transfer, use of Ll operations which have no equivalent in L2, etc.). Let us now return very briefly to the question addressed at the beginning of this paragraph: is transfer always a ‘knowledgeable operation’? It was argued that it need not be the case; but this again does not imply that the last mentioned constraint excludes (explicit or implicit) use of previously acquired knowledge about Ll and L2 structure. Zobl(1979b, 1980) discusses the importance of verb-complement cohesion and comes to the conclusion that it imposes restrictions on L2 development, including the use of transfer strategies. He interprets his findings-which receive independent support from Meisel (1980a)-as an argument in favor of an approach which focuses on structural properties of linguistic systems, as opposed to the learner asa creative hypothesis constructor. Such an opposition is, however, an artificial construct. Findings of research in language typology (e.g. the ‘hypothesis that consistent VO and OV languages seek to maintain an adjacency unity of verb and object’, Zobl, 1979b) must be related to cognitive principles if they are to be seen as operating in the individual’s processing of language. In the case mentioned, this has been done repeatedly, compare Slobin (1973, 1977). In other words, cohesion of linguistic subsystems is a relevant factor. This is not because of the linguist’s description as such, but because the learner processes language in this way. Consequently, the learner may or may not refer to it when perceiving two languages as typologically related. If he does, a number of other factors will intervene, as for example ‘naturalness’, as is argued by Vincent (1982) and ‘markedness’, as defined by Kellerman (1979).

TRANSFER AS A SECOND-LANGUAGE

STRATEGY

25

However, if the learner does not refer to such properties as cohesiveness in his intuitive evaluations of the two languages, they may nevertheless intervene in his L2 processing as habitualized psychological realities. 2.3. Use of transfer and otherstrategies by different types of learners 2.3.1. In our attempt to explain the occurrence of transfer in natural second language acquisition, a number of claims have been made about the nature of transfer and about which conditions must be met for transfer to occur. We have thus arrived at tentative answers to most of the questions raised above (see the introduction); the answers were deduced from the discussion of these questions in the psycholinguistic fr~ework outlined in 1.O. and from a review of that part of the relevant literature which does not merely offer a contrastive analysis of surface features. At this point, it seems necessary to tie together some of our hypotheses and to relate the use of transfer to other strategies of L2 acquisition and use. This discussion will be restricted to transfer as an instance of the acquirer’s hypothesis construction activity and will thus ignore some of the cases mentioned above, e.g. ‘non-knowledgeable’ application of performance routines. 2.3.2. Since we are working in a framework which views transfer as a mental activity, more precisely as one of a number of different strategies available to the learner, it may be useful to sketch briefly some aspects of the planning process underlying L2 speech production. The most appropriate way to do this, it seems to me, is to present a rough model in a flow chart as in Fig. 1. This model is obviously incomplete and does not aim at explanatory power. It is meant to illustrate which steps may possibly be taken on the Lt-speaker’s way from the intended message to the actual utterance and how they might be related to each other. Note that no claims concerning the speaker’s awareness are made for most of these steps. A person may or may not be aware of his lack of knowledge, to mention one example. In the former case, one can speak of ignorance-by-self-evaluation, as proposed by Kellerman (1977); this is, however, not a necessary condition for the learner to choose a solution different from the one suggested by the L2 grammar. Language production is obviously a compiex process of planning and execution, for the native speaker and for the L2 speaker who does have the necessary grammatical knowledge available as well; the L2 speaker will therefore always have to choose between different solutions, irrespective of how much knowledge about the L2 grammar he has already acquired (although this is not shown on Fig. 1). And he need not be aware of the fact that his solution lies on the path starting with insufficient knowledge. Biased, I believe, by the special learning situation he is studying (formal FL instruction), Kellerman (19771, in his model, only considers ‘self-ev~uation with respect to competence to express message in TL’. There are decisions which are specific to some learner types, and which very likely involve awareness. Some learners, for example, may check whether the solution they have chosen is an adequate one, and some will check whether they violate the target norm. These learners, in spite of knowledge already acquired, will search for different solutions as a consequence of their self-evaluation, In accordance with what has been said above (1.1.4), our model predicts that one can choose between two kinds of solutions. Either the function to be expressed is changed, or the formal device by which this function might be expressed is replaced by another one. A radical version of functional reduction would lead to silence; a more moderate one

26

JURGEN M. MEISEL Intended

message

Change

I Knowledge tb

of device

express

functm

A, X

IL-based

nypotheses

LI-based

hypotheses

.

SI

Utterance

Fig. 1. A

nce

modelof L2 speechproduction.

consists of changing the intended message. Both, however, entail loss of communicative potentialities. The other kind of solution involves replacement of an L2 structure by a structure which expresses a similar or identical function but which is generated without or with only partial knowledge of this part of the L2 grammar. This ‘interlanguage’-form A, may, on the surface, be identical to the LZform 4. It is arrived at by means of a ‘creative construction process’. Transfer is one such process, and it differs from the others in the kind of knowledge used and in the nature of the processes applied in constructing structure 4. As mentioned before, some but not all speakers will check whether the results of these processes conform to the requirements of the target norm. Note that this refers to the acquirer’s evaluation of grammaticality and adequacy. Whether the utterance produced does in fact obey the norm is an entirely different matter. If the learner perceives the structure as ungrammatical, he may go back to the point where he has to decide how to

TRANSFER

AS A SECOND-LANGUAGE

STRATEGY

27

change the utterance in form or in content. Similarly, if the planned utterance would be too difficult to execute because of a lack of habitualization of the necessary performance skills, the learner may go back to this point’of the planning process, even if the correct ~2 rules have been acquired; he is not always able to use his knowledge. Let us now have a closer look at the different solutions by which second language speakers aim at producing structures which are functionally equivalent to structural devices of the target variety not yet available at a given point in the development of L2 competence. Each of the boxes in Fig. 1 stands for a complex decision-making process. These processes are quite obviously similar in as much as they are constrained by restrictions on comprehensibility, expressibility and learnability which the structure of the mental system imposes on them, as on other psychological processes (compare section 1 above). Less trivially, they also have in common the fact that they are constrained by the amount and the kind of knowledge about L2 grammar that has been acquired at a certain point in time. This follows from a view of second language acquisition, shared by most scholars in the field, according to which language acquisition implies, among other things, the internalization of grammar by integrating new material into the previously acquired system, and only what can be accommodated to the already established system will be acquired and used productively. What is more interesting for the present discussion, however, is how the various solutions differ: they make use of different kinds of knowledge, and they also involve different operations. The description of the transfer strategy, as given in 2.2 above, should help to illustrate this point. This kind of transfer will only occur if the learner makes use of previously acquired L2 knowledge and of Ll knowledge; in addition, he must perceive the two languages as typologically close (i.e. contrast the two languages) and he has to assign a value of language specificity to the item to be transferred. If, on the other hand, the learner relies entirely on previously acquired L2 knowledge, he will arrive at different solutions, depending on which processes are applied. Either hypotheses can be constructed by merely extending some which are familiar from L2, or one can deduce new hypotheses by applying universal operating principles, implicit knowledge of Universal Grammar, etc. The former operation may result in what is usually called ‘overgeneralizations’. This view is in contradiction with Taylor (1975, p. 73): The overgeneralization and transfer learning strategies appear to be two distinctly different linguistic manifestations of one psychological process: reliance on prior learning to facilitate new learning.

Although in both cases reliance on prior learning is involved, overgeneralization is restricted to one source of knowledge (L2), and the processes are fundamentally different. What distinguishes extension of L2 hypotheses from the application of universal principles is that in the latter case second language acquisition will result in a developmental pattern which is less oriented towards the target system. These two alternatives correspond to the distinction between ‘acquisition towards an internal norm’ (nativization), and ‘acquisition towards an external norm’ (denativization/decreolization), as suggested by Andersen (1980). It is quite crucial to keep these two alternatives apart. For one thing, this accounts for the fact that L2 acquirers may very well undergo development, without, however, necessarily approaching the target norm. In other words, the hypotheses of the learner need not be solutions by which the expressiveness of speech is reduced, either with respect to function or to form; in fact, there is good empirical evidence that impoverishment of this kind is more likely to result from very strict obedience to ‘external norm’ than

28

JORGEN M. MEISEL

from ‘nativization’ (see 3.2). Secondly, in as much as the ‘internal norm’ is set by universal principles, similar solutions are found in natural second language acquisition, first language acquisition, creolization, historical change, etc. To sum up, one has to distinguish a number of processes which are frequently confounded; they are combined in different ways to constitute the various solutions available to the L2 acquirer. There is ‘expansion’ (and its counterpart ‘reduction’) which can be defined as suggested by Mtihlhausler (1980, p. 3) in his important paper on this topic: as comprising those additions or non-referential potential.

to an existing

linguistic

system that lead to an increase

in the referential

There is conformity to (or divergence from) the target norm, which depends on external factors as well as on the learner’s readiness to deduce his hypotheses from an analysis of the input rather than from universal principles. There is also previously acquired linguistic knowledge (first language, foreign languages, etc.) which if used, might lead to ‘language mixing’; see Miihlhausler (1980). And there is simplification-a tendency to keep the cognitive costs of language processing low by using structures which are performatively least complex-if necessary by subjecting them to changes. It should be noted that various processes of this kind-and the list given is certainly not complete-can be combined, whereas others do not occur simultaneously-as it happens or because they stand in logical contradiction. Thus, use of Ll knowledge may be preferred over other knowledge when it leads to a performatively less complex (more natural; compare Vincent, 1982) solution. A similar point is made by Mtihlhausler (1980, p. 12): I suspect that mixing is highly selective and that, even in the case of mixture of fully-developed systems, natural solutions are preferred. Nevertheless, in many instances mixing can lead to unnatural outcomes.

To mention another example, simplification need not entail violation of the target norm (compare Meisel 1980a, p. 35f), nor does it necessarily lead to structural reduction (compare Meisel, 1977; and Mtihlhausler, 1980). 2.3.3. To close this section, one should address the question as to what might influence a learner’s choice between the different solutions available. Our claim is that acquirers of a second language in a natural setting (and probably also foreign language students) develop different learning orientations. As was briefly mentioned in section 1, we support the claim that external factors, such as social setting, social distance from the L2 society, aspirations (intended length of stay in the L2 community, job opportunities, etc.) and so forth, together with individual factors like the personality of a learner (which, in turn, is partly also a result of social variables, e.g. primary socialization) merge into a network of social-psychological factors which determine specific kinds of attitudes and motivations. We want to regard these as different aspects of a person’s general orientation; in Meisel et al. (1981), where this approach is described in more detail, we suggested that these orientations can be plotted on a continuum ranging from segregative at one end to integrative at the other. These orientations influence very strongly the social behavior of an individual, including the specific approach to the language learning task, what we call the ‘learning orientation’. In Clahsen et al. (1983), we present a fairly comprehensive account of social psychological variables which constitute determining factors in natural second language acquisition, based on a cross-sectional analysis of the acquisition of

TRANSFER

AS A SECOND-LANGUAGE

STRATEGY

29

German by adult immigrant workers from Italy, Portugal and Spain. The linguistic characteristics of learner-type-specific interlanguage varieties are described, and an explanation for this variation is offered in terms of strategies of L2 acquisition and use. In a subsequent longitudin~ study, these hypotheses have been checked; in seven case studies we examine the relations between social-psychological factors, general orientations, learning orientations and the use of strategies. Special attention is paid to the role of attitudes, motivations and the learner’s personality profile-factors which were not easily accessible in the cross-sectional study. The results of this analysis will be presented in Clahsen et al. (forthcoming). At this point, I will merely sketch rather superficially, for the purpose of illustration, in what fashion learning orientations might determine the choice of specific strategies. As was already suggested in 1.1.4 above, learners can focus on structural or on functional aspects when coding an intended message. This should not be understood as a binary choice, however; both can be more or less expanded or reduced. Thus, a learner may expand both, i.e. strive for the expression of a great variety of semantic-pra~atic functions in as many ways as possible (the ideal learner); or he may tend to expand functional potentialities but reduce structural aspects, i.e. focus on the message, neglecting stylistic variation; and so forth. Add to this what has just been said about conformity with the target norm, use of previously acquired knowledge, etc.-see Fig. 1. These traits taken together will characterize learning orientations. Consequently, a learner is more likely to make use of those strategies which match his learning orientations. Taking the example of transfer from Ll again, all conditions on the application of the strategy mentioned in 2.2.1 and 2.2.2, with the exception of the first (lack of knowledge), point to the fact that this kind of transfer requires a structural learning orientation. This is a necessary feature of learning orientations favoring the use of transfer: there are, however, others which further stimulate this choice: in such learners functional orientation tends to be rather weak, and conformity with the target norm tends to be very strong. If the markings of the latter two features are reversed, the resulting orientation would give preference to the use of strategies based on universal principles leading to development towards an ‘internal norm’. Although reference to Ll knowledge is not excluded in this case, as long as structural orientation is also present, it is much less likely to occur. Although this description of learning orientations and their influence on the choice of strategies of language acquisition and use is merely tentative, it is hoped that it will at least seem plausible. A more satisfactory explanation would require an analysis of the origins of learning orientations in terms of personality profile, attitudes, etc., on the one hand, and a description of specific sets of linguistic features resulting from certain such orientations, on the other. Calling on plausibi~ty again, I would claim that a risk-avoiding person, self-controlled or controlled by others, oriented towards integration and so forth, is likely to adapt his behavior to the standards of the target group. With respect to language use, this means that in case of doubt, an intended message may be suppressed or changed to avoid risking errors. Those structures used will be analyzed with respect to formal properties, and monitoring will play an important role. This results, to mention some examples, in the use of obligatory rules of the target language even if they do not increase the expressive force of the utterance; optional rules, on the other hand, will be avoided even if this leads to monostylistic speech. For a comprehensive description of learner-type specific language use and a justification of the various learning orientations, see Clahsen et al. (forthcoming). The brief remarks made in this paper should, however,

30

JtiRGEN M. MEISEL

suffice to give an idea of how transfer from the first language could be related to the use of other strategies. 3. A review of some empirical evidence for Ll transfer The aim of this paper, up to this point, has been to take a deductive approach in our attempt to determine the role of Ll transfer as a strategy of L2 acquisition and use. We have thus arrived at a number of conditions on the application of transfer which enabled us to discriminate among various processes involved in the use of this strategy and to predict that certain types of learning orientations will favor the choice of Ll transfer, as opposed to simplification, overgeneralization, etc., in case of lack of knowledge or of performative skills. It was claimed that such an approach should lead to a more comprehensive understanding of the problem and could help to assess the actual role of transfer, even in those cases where different strategies produce the same structural results, making it thus impossible to decide, by means of an analysis of surface phenomena, whether transfer has occurred or not. The next step should obviously be to test the empirical validity of the hypotheses offered in the preceding sections. It may still be premature to take this step, because the list of conditions given in section 2 is clearly far from being exhaustive, and the learning orientations have only been defined very vaguely. Nevertheless, I will try to discuss some empirical findings insofar as they support or contradict these hypotheses, hoping that this discussion, in the long run, will lead to their revision and more adequate formulation. At the very least, I wish to appease those who dislike what may appear to be a speculative approach. In attempting to do this, I will proceed as follows: first, I will very briefly review which structural areas are claimed to be affected by Ll transfer; this will be limited to alleged cases of syntactic and morphological transfer, focusing on data from the acquisition of German by foreign workers. The aim of this review is to identify some good candidates for transfer, as defined in this paper. Secondly, two case studies of Spanish immigrants to Germany will be presented, and our predictions of possible occurrences of transfer are tested. The results, I hope, are amenable to falsification by other researchers. 3.1. A re-examination of some alleged cases of syntactic and morphological transfer In this part, those features of the process and those conditions on the application of transfer depending on the individual-e.g. learning orientation, development of L2 competence-are not considered. Rather, the aim is to find out which structural properties of Ll and L2 are most likely to be affected if a learner tends to make use of transfer at all. One would hope that this might eventually help to establish a hierarchy of transferability based on psychological markedness, as opposed to linguistic contrast, taking into account both language. A framework of this kind is suggested by Vincent (1982), and Keller-man (1977, 1979) proposes and tests criteria for the definition of language-neutrality and language-specificity for lexical material. In what follows, I will deal with four structural types of possible transfer: (1) omission of obligatory L2 surface elements which are optional or non-existent in Ll; (2) omission or use of inflectional markings in L2 if Ll lacks or possesses corresponding devices, (3) compound words and various other N-N-constructions modeled on the Ll pattern; (4) use of Ll word order in L2. 3.1.1. One of the most frequently mentioned cases of claimed (negative) transfer is failure

TRANSFER AS A SECOND-LANGUAGE

STRATEGY

31

consistently supply categories which are required in L2 but which are optional or lacking in Ll, e.g. subject pronouns, articles, copula/auxiliary, prepositions. I will argue that the case of missing elements is a bad argument for transfer: the evidence given so far at best allows the conclusion that if a device is non-existent in Ll, it may be less easily available for L2 use. This does not mean, however, that acquirers tend to form hypotheses about L2 which ignore the existence of this device in L2 grammar. The following arguments support, I believe, my claim. to

(a) As is admitted by many researchers, including some of those who believe that the omission of elements can be explained as negative transfer, e.g. Gilbert and Orlovii: (1975), Meyer-Ingwersen et al. (1977), St6lting (1980). For all the examples quoted, transfer is never the only possible explanation; all these cases could equally well be accounted for by means of simplification (‘pidginization’), overgeneralization, etc. In fact, simplified registers like foreigner talk, telegraphese, etc., do not differ from second language learners’ speech as far as missing elements are concerned; (compare Meise1,1977). Since these varieties of a language are spoken by monolingual native speakers, transfer must obviously be ruled out as a possible explanation. Do we, then, have to assume that in the case of simplified registers and similar varieties we are dealing with simplification, whereas in the case of L2 speakers, the same phenomenon is a result of a Ll transfer? Although this solution is logically possible, it has to be ruled out in the light of additional evidence. For one thing, learners whose first language does have these elements (e.g. articles, copula/auxiliary, prepositions in Romance languages; subject pronouns, articles, etc. in French and English) also omit them quite frequently when speaking a second language (e.g. German). Thus, interference would only account for a significantly higher frequency of omissions, whereas in the case of speakers whose Ll does have them, the omission of these elements remains unexplained by transfer, as well as important individual differences between speakers of the same Ll. But even the frequency argument is weakened by the fact that it only accounts for part of the data. Gilbert and OrloviC (1975) and Gilbert (1980) tested it for the deletion of definite articles, comparing Greek, Romance (Italian, Portuguese, Spanish), Yugoslavian and Turkish learners of German. Although they report that the latter two groups omitted the definite article more frequently than all the others, the differences between the Romance speakers are greater than between each of them and the Turks and Yugoslavs: hcurrence of definite article; from Gilbert and Orlovii (1975, p. 6) Turkish 15,91%; Portuguese 35,34%; Yugoslav 19,75%; Italian Spanish 87,10%.

It thus appears that transfer might, at best, merge with simplification and increase the number of omissions. But even this interpretation cannot easily be maintained, since a detailed analysis reveals that many, and for some learnerrs all omissions can be predicted for certain environments by means of simplification strategies, where transfer could only account for random deletions. Consider, for example, the omission of subject pronouns by Romance learners of German. I have tried to show elsewhere (Meisel, 1977, 1980a) that subject pronouns are omitted significantly more often in contexts where subject and verb would have had to be inverted. Moreover, the non-occurrence of pronouns can be predicted in cases where reference can

32

JtiRGEN

M. MEISEL

easily be established without the pronoun, either through the context (non-linguistic or linguistic) or by otherwise redundant marking, e.g. by verb inflection; (see (b) below). Related to this observation is the fact that third person pronouns are omitted most frequently and first person pronouns least frequently. These facts cannot be explained by the transfer hypothesis; a more extensive treatment of such data is presented in Clahsen et al. (forthcoming). (b) The remarks in (a) suggest that in a number of cases transfer cannot explain the omission of elements in L2 speech and that in all other cases it can merely be expected to reinforce other strategies employed by the learners. This conclusion is supported by an evaluation based on the results of section 2. Since nobody claims, as far as I know, that interference would lead to a structural feature being omitted categorically, its nonoccurrence cannot be interpreted as a failure on the part of the acquirer to perceive the existence of this feature in L2. Consequently, one could say, referring to the model of L2 speech planning (see Fig. 1) of 2.3.1, that elements are omitted because they represent a problem for the execution of planned speech, rather than being missing because of a lack of knowledge about L2 structure. Longitudinal data indicate, in fact, that frequencies of omissions do not necessarily show a developmental pattern; at least some learners omit subject pronouns, to take up this example again, approximately as often after more than two years of contact with L2 as they did after a few weeks, although they do make progress in many other aspects of L2 development; (compare Clahsen et al., forthcoming). This leads to a more general remark. It appears to me that merely focusing on the presence or absence of a given phenomenon is an inadequate approach to L2 acquisition in any case. In general, it may be assumed that if the Ll does not make use of a certain device to express a function, there will be some other device to do just this: demonstratives or inflectional marking to express definiteness (article), inflection on the verb to replace subject pronouns, noun inflection instead of prepositions, etc. The real question is then: how does the learner express what is usually coded by the missing element? Does he rely on the context or on the help of the interlocutor, or will he use some other device? To mention one example: Zita PL, a Portuguese learner of German who was studied longitudinally (see 3.2.), usually supplies verb inflection for the second person singular only in sentences where the subject pronoun is omitted, e.g. at one point there are twelve instances of inflectional marking in sentences without subjects, but only two with subject and inflectional marking supplied. The reverse apparently also holds true: omission of subjects entails verbal marking, with only seven exceptions during a period of more than one year (at least one recording analyzed per month). A very interesting observation is reported by Stolting (1980) who studied the use of German and Serbocroatian by Yugoslav children in Germany. Among the very few cases of possible syntactic transfer from Serbocroation to German found in these data, there is omission of articles and of subject pronouns. The frequencies, however, are very low: omission of articles ranging from 0.13 to 0.04, omission of subject pronouns 0.03 to 0.0 (mean group scores). Interestingly enough, Stolting observed an increase in the frequency of demonstratives in the Serbocroatian speech of the same children, e.g. the device which corresponds roughly to the German definite article when used anaphorically, as is in fact the case in the data. Although a similar increase in the occurrence of Serbocratian (optional) subject pronouns could not be found, these findings suggest that transfer may be more likely to go from L2 to Ll. This phenomenon could be explained by a tendency,

TRANSFER

AS A SECOND-LANGUAGE

known from Ll acquisition and from creolization, and clearly’ (Slobin 1973).

STRATEGY

33

to mark a semantic function ‘overtly

(c) A final point concerning omitted elements is that the empirical findings by different researchers are not consistent. Meyer-Ingwersen et al. (1977) quite generally assume in a rather simplistic way that transfer plays a crucial role in L2 acquisition and stress the importance of missing articles, copulas, subject pronouns and prepositions in the speech of Turkish children. Steinmtiller (1981), on the other hand, reports in a case study of a Turkish child that for articles and pronouns he showed a high level of proficiency; and Pfaff (1979a) and Pfaff and Portz (1979) who studied the acquisition of German by, among others, Turkish, Greek and Yugoslavian children, found very few instances of the omission of the copula. Similar incompatibilities show up with respect to the chronology of the appearance of such phenomena: Meyer-Ingwersen et al. (1977) claim that transfer predominantly occurs during later learning phases, whereas Pfaff (1979) and Stolting (1980) report that omissions are found during early phases, if at all. 3.1.2. There seems to be general agreement among scholars studying L2 acquisition that free morphology can come from the first language but bound morphology cannot be transferred; (compare Krashen, 1977b). A similar point is made in 2.2.3 above, referring to internal cohesion of constituents; in addition, findings from code-switching by bilinguals corroborated this claim, as is evidenced by the ‘free morpheme constraint’ (Poplack, 1980) or by Pfaff (1979b, p. 303): Clitic pronoun objects are realized in the same language as the verb to which they are cliticized, and in the position required by the syntactic rules of that language.

Clitics have to be considered as belonging to bound morphology; this is confirmed by their syntactic behavior in general, e.g. no non-clitic constituent can go between the verb and its clitic. From these remarks it follows that specific forms of Ll inflectional morphology cannot be transferred to L2; learners appear to judge them intuitively as highly ‘language-specific’ (Kellerman 1977). Note that Schumann ( 1978), supported by Andersen ( 1979), contradicts this assumption by claiming that ‘positive transfer from Spanish can account for the plural inflection being supplied 85% of the time’ on nouns by Alberto, the learner studied by Schumann; i.e. transfer is stimulated by phonetic and semantic similarity of this morpheme in both languages. Additional evidence for this interpretation is not offered, however; it certainly conflicts with the assumptions just stated. Empirical findings by Steinmiiller (1981) who reports that nominal plurality in German is the most difficult feature to acquire of those studied in the language of his Turkish learner, point to a different interpretation of the hypothesis about the transferability of inflectional morphology. This ‘weak version’ would predict that positive transfer might occur if Ll and L2 possess devices expressing the same function, irrespective of their phonetic form. The lack of a functionally equivalent form in Ll, on the other hand, should make the acquisition of the L2 device more difficult. Note that this is a variant of the claim concerning omission of elements, discussed in 3.1.1. Consequently, the counterarguments given above also apply here. It therefore does not come as a surprise that the empirical evidence is not convincing. Schumann (1978, p. 65f) resorts to positive transfer to account for is (cop)‘s being correctly supplied to a greater extent than other auxiliaries and for am (cop), and can reaching criterion for appearance

(cop), are

34

JfiJRGEN

M. MEISEL

in Alberto’s speech. Andersen (1979), reinterpreting Schumann’s findings, goes even further and adds the progressive -ing inflection to the list. It seems that Albert0 did not really learn English syntax and morphology; all cases which may appear to constitute counterevidence to this claim are eliminated as positive transfer. This line of reasoning is clearly in contradiction with our definition of transfer: When his knowledge of English, especially English syntax, is lacking, he productively relies on a linguistic system he already commands--Spanish.

(Andersen, 1979, p. 12)

In the light of the hypotheses discussed in section 2, such a statement implausible, especially since it is not given further argumentative support.

appears to be

Furthermore, numerous studies have shown convincingly that the copula and some medals are acquired during early phases of L2 acquisition (compare, for example, Felix, 1978; Pienemann, 1981) and that irregular verb forms are also acquired rather early, although not necessarily functionally equivalent to the L2 norm; for some recent discussion compare Meisel (forthcoming) and Wode (forthcoming). This conformity cannot be accounted for by means of the transfer hypothesis, given that a variety of source and of target languages have been investigated, Pfaff and Portz (1979, p. lff), also write that the majority of the features investigated, word order, irregular verb forms, person/number agreement, generalization of the auxiliary, appear to develop similarly in Turkish and Greek learners despite considerable typological differences in the first languages.

Similarly, Connors (1980) found that errors in the marking of irregular verbs made by Canadian immersion pupils (Ll English, L2 French) could not be attributed to Ll interference but to LZ-based learner hypotheses (using vais as the second person singular). A last example should show that even in its weakest form-facilitation of L2 production if a corresponding feature exists in Ll, complication if Ll lacks this feature-the transfer hypothesis is not a very useful explanation of development of inflectional morphology. Gilbert and Orlovic (1975) and Gilbert (1980), arguing in favor of the transfer hypothesis, had to admit that it cannot predict problems with case marking on the German definite article. There is a strong tendency for learners to use an invariant form die, which is the most frequent one (Gilbert calculates a type ratio of 44% and a token frequency of estimated 50%); die replaces other forms in some German dialects as well. This tendency is documented in a number of different studies. Gilbert (1980, p. 8) states: Strict adherence to the transfer hypothesis would lead to the expectation that speakers of those source languages having inflectional categories most similar to those of German would show more non-die forms... This should place the six languages in the rank order: Greek; Italian-Spanish-Portuguese; Serbo-Croatian-Turkish... Instead it appears (aside from the Jugoslavs) that there is a negative relationship between increased use of the definite article and increased use of the marked forms (der, das, dem, den, des).

There are, however, two cases of possible transfer, reported by Pfaff and Portz (1979), which apparently support the claim that free morphemes may indeed be affected by Ll transfer. German allows indirect object marking either in ‘analytic form‘ as inflectional marking on determiners, pronouns, etc. or by means of a preposition, plus inflectional marking. Learners tend to use prepositions in cases where standard German only allows inflectional case marking. Pfaff and Portz found that Greeks made more inappropriate use of prepositions (e.g. das Miidchen gibt zu dem Jungen, instead of das Miidchen gibt dem Jungen ‘the girl gives (to) the boy’) than the Turkish children; the former used zu

TRANSFER

AS A SECOND-LANGUAGE

STRATEGY

35

‘to’, in ‘in’ and bei ‘at’, whereas the latter only used zu; and, finally, the Greek speakers conformed to the Greek co-occurrence restrictions (prepositions preceding full nouns, pronominal objects marked inflectionally). The second case refers to gender marking. Greek, like German but unlike Turkish, has gender marking. Pfaff and Portz found that when grammatical and natural gender coincide (German: die (fern) Tochter ‘the daughter’, but dus (neuter) Msidchen ‘the girl’), the Greek speakers marked subject pronouns, possessive pronouns and definite articles more consistently for gender than Turkish learners. Pfaff and Portz (1979a, p. 20) summarize .. . it appears that structures which are realized primarily by lexical morphemes, e.g. the analytic dative (realized with a preposition) or gender (realized with such distinct personal pronouns as er vs. sie or sein vs. ihr) are more subject to interference than structures primarily realized by means of general inflection or word order rules.

Unfortunately, the facts are not quite so easily explained, although this interpretation does predict that case and gender marking on determiners should not be affected by positive transfer, as was in fact found by Gilbert (1980). It should be added, however, that gender marking, when restricted to natural gender assignment, is basically a semantic stragegy. This is supported by the observation that speakers of source languages which do possess grammatical gender tend to transfer gender assignment from semantic equivalents in Ll . This is well-known, and a list of such examples is given by Stolting (1980). Similarly, I would claim that the inappropriate use of prepositions to mark indirect objects is a semantically based strategy. In fact, these prepositions express the beneficiary of an action, rather than the indirect object. Similar examples as those quoted by Pfaff and Portz (1979) are also documented for other Ll groups and with other prepositions, e.g. ich sage fiir dir ‘I say for you-I’ll tell you’ (Ll Spanish). Note that Steinmttller (1981) also found in with a Turkish learner, expressing all kinds of relations; whereas Pfaff and Portz only mention zu appearing in the Turkish data. In sum, here again two processes appear to be involved. On the one hand, both features seem to result from semantically based strategies, expressing natural gender or underlying semantic relations (e.g. beneficiary). But apparently those learners who are used to expressing these semantic functions in their first language are more likely to make use of free morphemes, and they tend to prefer a device which resembles those familiar from the first language. Thus, as in the case of omissions, discussed in 3.1.1, transfer can be said to operate in such a way that reliance on prior knowledge is likely to support the choice of certain solutions and stimulate the use of specific devices. 3.1.3. The third case deals with constructions by which nouns are put together, as in possessive NP’s NP. Andersen (1978) gives a detailed discussion of these structures as they appear in the speech of Spanish (Ll) learners of English. He found three forms in his data (1) NsN (2) N N (3) NofN

my uncle’s house my uncle house the house of my uncle

native (L2) form partial native form Spanish-based form

The third example corresponds to Spanish la coscl de mi tio with the modifier following the head noun. Example (2), frequent in English Ll child language, is based on L2 knowledge but the inflectional marking is missing. A fourth possibility also occurred in Andersen’s data: (4) N N

house uncle

36

JtiRGEN

M. MEISEL

This results if Spanish word order is retained without inserting the English equivalent (ofi of Spanish de. Andersen, furthermore, points to the similarity in surface form of the incorrect case (2) to English noun adjunct constructions (NN) as in (5): (5) the brick house a cement wall, etc. (6) the rest of the day, a bottle of milk

(6) gives examples of English N ofN constructions. The Spanish equivalents of both (5) and (6) are like that for (1): N de N. In all three cases Andersen finds, in fact, N of N structures in the speech of Spanish learners. ‘Thus, the transfer strategy is not confined to possessive constructions.’ (Andersen 1978, p. 4). The fact that most of the learners studied used at least two of the three forms given in (l)-(3) indicates that transfer is only one of several possibilities for the learner. Some learners only produced forms (1) and (2), thus not making use of Ll knowledge; and some used (3) together with either (1) or (2) or both, thus using Ll only occasionally. Only two of the 89 subjects studied by Andersen relied entirely on the third possibility. In phrasing it this way, I am not following Andersen who believes that the use of the different strategies is chronologically ordered: transfer during early phases of L2 acquisition, later on transfer plus L2 based learner hypotheses (‘overgeneralizations’, etc.), finally only the latter. As stated before, I argue that Ll transfer is primarily a matter of different types of learners, and only for those who use it at all can the question of chronological ordering be raised. Andersen’s main point, however, seems to be well established: there are errors which can only be explained by an interaction of both first-language transfer and ‘second-language overgeneralization’ in the case of these noun phrase constructions. If this is correct, it involves transfer of Ll word order as well as of free morphemes(de-, d). Andersen (1979) points out that Alberto, the learner studied by Schumann (1978), used forms like (4) above in 43% of his possessive constructions: e.g. the arm the girl instead of the girl’s arm; this ‘pidginized’ learner thus mainly relies on Ll word order. Let us now look at some data from learners of German. In German, there is a construction similar to the English possessive structure (see 1): (7) meines Onkels Haus

‘my uncle’s house’

I believe, however, that this can only be used in extremely formal or even bookish registers. Rather, one would have to say: (8) das Haus meines Onkels

i.e. the head noun precedes, the modifier noun (marked as genitive) follows. In colloquial spoken German, on the other hand, constructions like (9) are very frequent: (9) das Haus von meinem Onkel

This parallels the English interlanguage structure as in (3). the one which Andersen explains as a result of Ll transfer. Given that an immigrant acquiring German in a natural environment, or even an immigrant child going to a German school, will never hear sentences constructed like (7), data from Spanish learners of German seem to be uninteresting for the current discussion. The German target variety uses the same word order as Spanish, and if there is a preposition, it is just the one which corresponds to Spanish de,

TRANSFER

AS A SECOND-LANGUAGE

STRATEGY

31

German von. Two exceptions have to be added: first, N s N structures are used in colloquial German if the modifier noun is a name, as in (10); (10) Peter Haus

‘Peter’s house’

second, dialectal varieties of German allow constructions noun as dative and postponing the possessive pronoun: (11) dem Onkelsein Hause (12) meinem Onkel sein Haus

Like (1 l), marking the modifier

‘(the) uncle’s house’ ‘my uncle’s house’

Nevertheless, since colloquial German offers all these possibilities @N s, -N von N, N and even Ns_N-head noun underlined), it will hardly be possible to determine Ll influence in the case of Spanish learners. Only constructions like (5) could be interesting, for German requires compound constructions in these cases where Spanish has N de N

possfi,

(13) das Backsteinhaus, eine Zemenfwand, etc.

As far as possessive structures are concerned, it appears to be more promising to look at Turkish learners. Turkish obligatorily requires the order N + GENITIVE + head noun, Meyer-Ingwersen et al. (1977, p. 189ff) have found that Turkish children exclusively use structures where the modifier noun precedes the head noun: (14) (15) (16) (17)

Jung Vater der Junge sein Papa Kind sein Voter kleine Junges Voter

‘boy father’ ‘the boy’s father’ ‘the boy POSS daddy’ ‘child POSS father’ ‘little boy + GENETIVE father’

Although there is a possible German model for each of these sentences, it is surprising that the more frequent preposing of the head noun never occurs, nor the construction with von, which is probably the most frequent one in colloquial German. The Turkish children seem to pick up constructions like (11) and (12), and Meyer-Ingwersen et al. explain that this is the one which corresponds most closely to the Turkish structure. Pfaff (1979a) also reports that Turkish children use postponed head nouns in expressing part/whole and possessive relationships. She has also found N von Nconstructions, which is not surprising if one keeps in mind that not all learners will use Ll transfer and that those who do need not use it equally often at all points of L2 development. Pfaff even gives further evidence for the transfer hypothesis by stating that none of the speakers of the native German control group (in Berlin) used sentences like (11) and (12), i.e. those which supposedly are the models for the Turkish learners. To sum up, then, possessive constructions seem to be affected by Ll transfer, at least as far as word order is concerned and in the case of at least some types of learners. Again, transfer appears to be enhanced if similar forms also exist in the learner’s input; furthermore, the resulting construction is characterized by analytical marking (Pfaff, 1979a) and may thus be the cognitively less complex solution, as compared to constructions using inflectional. case marking. In other words, possessive constructions represent a good candidate for a structural area where Ll transfer may be a useful strategy for the learner, especially if it leads to a non-complex (‘unmarked’) case, and if it is favored by similar L2 patterns in the input. 3.1.4. The last point deals with patterns following Ll word order. This can be brief since

38

JORGEN M. MEISEL

some examples have already been mentioned above (compare 2.2.3 and 3.1.3) and in various other papers, e.g. Meisel (1980a,b). The arguments then given concerning Ll transfer of word order phenomena will be maintained, but a few additional observations can be made. Basically, this amounts to saying that L2 acquirers are using the underlying canonical word order of Ll as a starting point in L2 acquisition and in speech processing in the second language. They tend to retain this word order, especially if L2 input confirms the initial hypothesis; this includes cases in which the second language differs in environmental restrictions for a certain operation, as compared to Ll , i.e. overgeneralization of Ll structures in the second language. Even for word order phenomena, however, transfer is only used by some types of learners and it is more likely to occur if the resulting structure is easier to process than possible alternatives offered by the target variety. More specifically, this position predicts that underlying SVO or SOV word order of Romance and Turkish learners, respectively, will tend to appear in German interlanguages. Clahsen (in press) gives strong empirical evidence for this claim, analyzing the speech of Romance speaking learners of German. Meyer-Ingwersen et al. (1977) report that Turkish children frequently use verb final constructions where German would require SVO but, unfortunately, they never quantify their data and they do not give an unambiguous example either. Pfaff and Portz (1979), on the other hand, state that in their study Turkish children use verb final structures less frequently than Greek learners. Further empirical evidence is needed to clarify this point. But there is such evidence from studies dealing with other languages, e.g. Bickerton (1977) and Bickerton and Odo (1976) report that Japanese speakers of Hawaiian Pidgin English prefer verb final constructions, and Schumann (1978) and Andersen (1979) mention retention of Spanish word order in Alberto’s English. A number of other cases of possible Ll transfer can be predicted, cases related to the one just mentioned; namely, whenever a noun phrase is modified, learners may transfer the order of head noun and modifier. This occurs in the possessive constructions discussed in 3.1.3, but it may be expected to appear in other instances as well, e.g. noun-adjective sequences and relatives. The latter example is discussed extensively by Meyer-Ingwersen et al. (1977), contrasting German and Turkish. Transfer of operations changing the underlying canonical word order is not well attested; I want to argue that transfer here plays a minimal role. A number of previous publications (compare Meisel et al., 1981, Meisel, 1980a; Clahsen, in press) offer abundant evidence that Italian, Portuguese and Spanish acquirers of German master some structures during early phases of L2 development which differ maximally from their first language (e.g. moving the non-tensed part of the verb into clause final position in main clauses), whereas other structures are used rather late, in spite of being almost identical to Ll patterns (e.g. adverbials in intrasentential position, between the verb and its object NP). There is only one case where one might suspect that positive transfer is involved: the inversion of subject and verb. German obligatorily requires this change of word order in a number of syntactic environments in which the three Romance languages allow the canonical SVO order, e.g. after topicalized noun phrases or adverbials. Only in interrogatives do we find similar although not identical conditions on the applicability of the inversion rule. Nevertheless, inversion is acquired rather late, and most learners use it only occasionally after they have acquired it; furthermore, as was shown in Meisel(1980a), these learners do not necessarily apply it earlier or more frequently in interrogatives than in other contexts. Still, one might

TRANSFER

AS A SECOND-LANGUAGE

STRATEGY

39

suspect that those who predominantly use inversion in interrogatives are the ones who do resort to transfer from Ll . Yet, as is demonstrated in Clahsen (in press), the subject often appears postverbally and sentence-finally, even when Standard German would require sentence-internal position of the subject. One could say that . . . most of the apparent applications of INVERSION . . can simply be described by extraposing the subject to sentence-final position.

From this it follows that even for these learners transfer does not appear to be a satisfying explanation of word order changes. It can, at best, be used to account for the fact that learners feel the necessity to express interrogation by means of syntactic changes, although I would tend to explain this differently: inversion in interrogatives serves a pragmatic function whereas in other contexts, e.g. after topicalization, it is solely a syntactic device required by the target grammar. Finally, I want to argue that negative constructions in the speech of L2 learners also support the claim that Ll transfer is of minimal importance for word order phenomena resulting from movement operations. Negation is probably the part of grammar which has been studied most thoroughly over the last decade, as far as studies of second language acquisition are concerned; see Wode (198 1) for the most comprehensive analysis to date. There seems to be a consensus that for a large variety of target languages and an even greater number of source languages, learners first go through a phase of preverbal negation before they use postverbal negation, even when acquiring languages which obligatorily require postverbal placement of the negator in case of propositional negation. Some researchers, e.g. Bickerton (1977), have claimed that preverbal negation results from Ll transfer. This does not explain, however, the fact that the same constructions appear in the speech of learners who do not have preverbal negation in their first language. Similarly, the placement of negators after auxiliaries and modals is left unexplained. Learners of German, for example, during a phase when they never place the negative element after the verb, consistently use Mod/Aux + NEG + V constructions. With simple verbs, German requires, leaving details aside, clause-final placement of the negator; in Mod/Aux + V structures, on the other hand, German requires NEG-placement between the modal or auxiliary and the verb. It thus appears as if the learner had already acquired the more complex case but is still incapable of negating a simple verb correctly. I will not go any further into this question since I dealt with it in some detail in Meisel (1980a). I argued that a semantic-syntactic strategy could adequately describe and explain the facts; this strategy requires that the negator should be placed immediately before the element to be negated. Clahsen (in press) suggests that it might be more adequate to state that the negator is placed immediately next to the verb, for it appears that some learners do not go through a phase of preverbal negation; instead, they place NEG immediately before or after the verb, but not after the complement, as required by the target grammar. Findings in our longitudinal study (see below, 3.2) indicate that this may be the case if a learner is oriented towards conformity with the target norm (compare 2.3.2). At any rate, neither preverbal negation nor placement of the negator immediately after the verb can be explained by transfer from Ll . To sum up this discussion, the empirical evidence available suggests that underlying word order and free morphemes are most likely to be transferred from first to second language. This confirms the hypothesis deduced from the above stated assumptions about the nature of transfer as a mental activity (1.1) making use of mental representations of linguistic

40

JORGEN M. MEISEL

knowledge and of habitualized operations of Ll processing (2.2.3). It should also be noted that semantic factors are involved in most if not all instances of syntactic and morphological transfer. The examples of free morphemes showed this quite clearly, and they could even be compared to lexical transfer. Avoiding rearrangements of underlying word order, on the other hand, not only preserves syntactic constituents but also guarantees the integrity of semantic units. In addition, transfer is much more likely to occur if there is convergence with other strategies, e.g. simplification or overgeneralization. The existence of L2 structures which at least in part confirm the learner’s Ll based hypotheses further stimulates the use of transfer strategies; this might have been expected in view of what has been said in 2.2.1 about partial knowledge of L2 as a necessary condition for transfer. Finally, transfer from the first language can be said to play a much more crucial role as a strategy of second language use than as a strategy of L2 acquisition. There is little evidence that prior knowledge might facilitate the forming of hypotheses about L2 in a substantial way, and even less that the lack of a certain device in Ll might seriously inhibit its discovery in L2. 3.2. A glance at two case studies The arguments developed in this paper are derived from experiences with studies of natural second language acquisition by our research group ZISA (Zweitspracherwerb italienischer und spanischer Arbeiter) see references listed under Clahsen and Meisel. The main body of work consists of a combined cross-sectional and longitudinal study; the former is presented in Clahsen et al. (1983), the results of the latter will be published in Clahsen et al. (forthcoming). I In the longitudinal study, twelve immigrants from Italy, Portugal and Spain to Germany have been interviewed for at least 57 weeks, some for well over 80 weeks, usually starting within a few weeks after immigration to Germany. At the time of the first interview they were between 14 and 37 yr of age. The data consist mainly of informal interviews, free conversations and some additional tests for language proficiency, linguistic awareness, etc. Comprehensive accounts of data collection procedures and analyzing techniques are given in the two publications just mentioned. Brief descriptions are included in Meisel (1980a) and Clahsen (in press). The linguistic analysis attempts to describe the development of the second language, German, focusing on syntactic and morphological aspects. We also tried to distinguish between developmental patterns which can be predicted for all learners, and others which are claimed to be specific to certain types of learners (see 1.1.4 above). In an effort to explain learner-type specific linguistic behavior by means of external and internal factors (compare 1.1.4 and 2.3.2), both the cross-sectional as well as the longitudinal study included an analysis of social-psychological factors, in the first case by means of a computer-aided statistical treatment. In the longitudinal study, individual factors were studied with special care. For that reason, case studies for each of the twelve learners were prepared which contained demographic information and reports about the individual learner’s contact at work, activities after working hours, etc. In addition to that, we carried out IQ and personality tests and administered an assertion analysis to determine motivations and attitudes. We believe that these analyses enable us to determine the learning orientation (see 2.3.1) of each individual resulting from the network of social-psychological factors, and we also think that we can show how learning orientations influence the choice of a person between

TRANSFER

AS A SECOND-LANGUAGE

STRATEGY

41

different strategies of L2 acquisition and L2 processing. To conclude this paper, I will briefly mention two of the learners studied who represent two different kinds of learning orientations. Whereas Ana is focusing on structural aspects of language and is anxious to conform to the linguistic norms of the target language, Jose is much less concerned about structural aspects and even less about conformity with the target norm, at least not if this would restrict the range of functional potentialities. I will give a few examples from the speech of both learners to illustrate that it is in fact the type of learner represented by Ana who is more likely to make use of Ll transfer. Ana was 22 years of age when she came to Germany. Until then she had lived in Andalucia, most of the time with her parents. She did not finish high school, studied interior decoration but did not continue until graduation, either. She went to Dusseldorf (Germany) to live with her German boy-friend whom she had met in Spain. There she lived rather isolated, as she did not have any contacts with neighbors; she did not make friends, because she hoped to leave Germany as soon as possible. A. took casual work as a cleaning lady, but virtually the only person with whom she communicated more often than occasionally during her stay in Germany was her German friend. Since he knew Spanish very well, she only spoke German to friends of her friend who visited, to the interviewers, and occasionally at work or when she went shopping, etc. During the first months of her stay, A. sometimes went to a Spanish club, and she also took evening classes to learn German (one evening per week), but she gave up both activities after a few months. Although she met Germans through her friend, she complained that she could not participate in conversations; on the other hand, she did not try actively to become involved, and she did not want to speak German when one of the interlocutors knew Spanish; she tended to use others as interpreters in these situations. This did not change significantly with A.‘s increasing knowledge of German. She felt rather insecure and underestimated her ability to communicate in the second language; she frequently made self-corrections and commented on her own use of German. This was reinforced by her friend who corrected her whenever she used German in his presence. After 65 weeks of stay in Germany, A.‘s relationship with her friend broke up and she went back to Spain. There she took German lessons again and met her teacher after classes. Her own impression, as well as the interviewer’s was that it was now easier for her to communicate in German, once there was no pressure on her any more to be a successful learner. Jose was 16 years of age when he came to Germany. He had lived in a village in Galicia, near the Portuguese border; for seven years, he had been living separated from his parents who had emigrated to Germany. He finally followed them because he could not find a job in Spain. During his first year of residence in Germany, he did not come into contact with Germans; since he did not have a working permit, he spent most of his time looking after his younger sister. One year after his arrival, J. started working in a Spanish restaurant in Dusseldorf; this was when the first recording was made (53rd week). Within one year, he changed his employment twice, mostly because he did not get along with his employers, but also because he could get better pay. During the first year of interviewing, he hesitated about whether he should go back to Spain or stay in Germany. In the meantime, he has settled down, feels ‘established’, and no longer thinks of going back. Nevertheless, he does not want to integrate into German society; his friends are Spanish, Portuguese or Italian. On the other hand, he enjoys his job, in spite of unfavorable working hours which leave him with practically no leisure time. He knows that it is useful for him to speak German, and he himself explained that tips increase if one chats with the customers. This

42

JijRGEN

M. MEISEL

and German girls were his main motivations to learn German. Jose never attended German language classes, and he never thought he might need this. He is so selfconfident that he was always sure to be able to pick up enough in. every-day conversations. He regarded this as a matter of intelligence and made derogatory remarks about his father who had only acquired some basic knowledge of German in more than ten years. In a number of publications, e.g. Meisel et al. (1981), Clahsen ef al. (1983), we have presented and justified various criteria which can be applied to determine the development of German as a second language; among these, the acquisition of certain word order phenomena is claimed to be prime indicator of the level of L2 development. According to these criteria, both Ana and Jose belong to the group of learners who make more progress than most other individuals studied by our research group. Jose reaches the last developmental stage thus defined, and Ana, although much slower, finally arrives at only one stage below Jose; see Clahsen (in press) for quantification of relevant syntactic phenomena. Other syntactic and morphological features, however, indicate that Ana and Jose take different approaches to the learning task. During early phases, Ana’s utterances are comparatively complex, even in cases where this structural expansion does not necessarily lead to an increase in functional potentialities. Quite to the contrary, it seems to force her to concentrate on the construction of complicated patterns and gives the native interlocutor the impression of hesitant speech. Jose, on the other hand, reduces his utterances sometimes quite drastically at the beginning; only after he has acquired the necessary knowledge and the skills to perform certain structures does he expand more. In fact, his developmental pattern shows a steady increase in structural complexity, whereas Ana appears to ‘fossilize’ fairly early. Native speakers rate Jose’s speech as more fluent and as closer to the target variety than Ana’s, even at times when an analysis reveals that J.‘s utterances are not less ungrammatical than A.‘s. A few examples might illustrate these points. J., during the first months of interviewing, reduces his utterances by omitting elements; up to 75% utterances per interview consist of one constituent only; for A. the corresponding figure is 44%. An additional feature to be considered is the expansion of the verb phrase and the number of adverbials, contained in the verb phrase or following as sentential adverbials. J. initially uses only V + PP structures (where PP represents all kinds of adverbials). We found that, in general, learners first use V + PP, then V + NP, and then gradually add more constituents. A. very early uses V + PP, V + NP, and V + NP + PP, she is the only learner of those studied by our research group who displayed such a variety at this point of development. However, she never gets far beyond this point, e.g. there are no utterances with more than two complements-objects or adverbials-in her speech. J. expands much more during later phases; he even uses V + PP + PP + PP, and he is one of the very few learners to use utterances with two objects fairly frequently: V + NP + NP. Similarly, A. very early produces embedded sentences. And she is the only learner in our longitudinal study who refers to future events/actions by means of the Standard German analytical future, we&en + V + INFINITIVE, whereas others use an adverbial like morgen ‘tomorrow’ and the verb in the present tense or not inflected. Referring to past events and actions, J. and other learners who are more oriented towards functional gain and less towards structural expansion, tend to omit the auxiliary, putting the remaining main verb in the past participle, usually ge + V + t (e.g. gesagt ‘said’); frequently, this verb is even moved to

TRANSFER

AS A SECOND-LANGUAGE

STRATEGY

43

the end of the clause, as required in German. There is no functional loss, reference to the past is successfully established but structural problems are avoided, e.g. tense marking on the auxiliary. Ana never does this. All this amounts to saying that A. is focusing on aspects of language form and on conformity with the target norm; she almost behaves like a tutored learner, as a consequence of her self-control, and possibly reinforcement by the behavior of her German friend. J. puts more emphasis on function; this is not to say that he neglects the structural part of the learning task, but he would not sacrifice functional potentialities to structural adequateness of utterances. According to our predictions stated above, learners of the type like Jose are not likely to use transfer strategies, whereas learners like Ana can be expected to use them more readily. These predictions appear to be borne out by the empirical evidence available. Jose sticks to the canonical underlying SVX order, as all our learners do to some extent, but this has been shown to be an indication of reduction of performative complexity, rather than of Ll transfer. Similarly, he almost never fails to apply subject-verb inversion in interrogatives, but he also uses it in contexts where neither Ll nor L2 would allow it; thus pragmatic factors (see above) and other ‘approximative hypotheses’ can account for these phenomena. Also, J. frequently omits subject pronouns, but only starting in the 95th week, which cannot be explained by transfer; rather, this seems to be an indication of stylistic flexibility, for it occurs preferably in narratives where similar structures are also used in the target variety, e.g. hab gesagt er self gehen ‘(I) ‘ve told him to leave’. Negative constructions never follow the Ll model in Jose’s speech. Adjective-noun constructions always obey the target norm. Interestingly enough, he even does this when he uses Spanish lexical items, e.g. die mariscos is ales von Galicia pescador ‘the seafood all comes from Galician fishermen’. Similarly, in German compound words which correspond to the Spanish NdeN-type (compare 3.1.3), Jose follows the German order. Even in cases with Spanish lexical material, e.g. meachaniker de caches, automechaniker, he first mixes Spanish and German, looking for the right word but uses the German construction as soon as he has found the appropriate German word. Very rarely, there are examples which seem to be influenced by Spanish-I have found just two utterances, but there may be a few more: de ball von tennis ‘the ball of tennis’ and de fiinf von dezember ich hatte achtzehnjahr ‘the five of December I had 18 years’. Ana also sticks to the canonical underlying SVX order for a rather long period of time, approximately one year after immigration. And she uses subject-verb inversion predominantly in interrogatives; 70% of the applications of this operation occur in this context. As argued before, these cases, however, do not seem to support the claim that Ll transfer is used, at least not without more convincing evidence. Such evidence can be found in Ana’s speech, as opposed to Jose’s use of German. Ana frequently uses noun-adjective constructions in the Spanish order: N + Adj, where German requires Adj + N order, e.g. eine haus que is eine haus alt ‘a house que is a house old’. During the first 40 weeks there seems to be no exception to this order: later on both orders, preposed and postposed adjectives, occur in A.‘s utterances. Given that she avoids all risks of errors if possible, it is not surprising that few compounds occur. Constructions like those used by Jose, e.g. kleineprivatalbum ‘little personal album’, cannot be found. The new constructions which would require a compound in German tend to follow the Spanish model, except that Ana omits the Spanish preposition de or its German equivalent von, e.g. tag fest, instead of festtag or Feiertag modelled on dia de fiesta. She also appears to transfer free morphemes,

44

JORGEN M. MEISEL

e.g. heiraten mit in contexts where German would require heiraten ‘to marry’. Similarly, Ana uses the complementizer para ‘for to’ in German, a translation which corresponds to the Spanish preposition para, e.g. kein geld fiir kaufen biicher ‘no money to buy books’, vorige woche St. spreche mit Iehrin fiir wissen wieviel lehrer in aula ‘last week, St. talk to teacher to know (find out) how many teachers were in the classroom’. German requires urn zu in these cases; the preposition fiir makes it even difficult to comprehend these sentences. These instances of possible Ll transfer are not very numerous, but they would suffice to support the claim that certain types of learners do resort to transfer as a strategy of second language use. In natural second language acquisition, this is, nevertheless, only one of a number of explanations for specific aspects of interlanguages. In most cases, convergence of strategies, apparently, is the most adequate explanation. As in real life, linguistic phenomena can hardly have only one parent. And, as I have tried to show, it is extremely difficult to give convincing empirical evidence for what caused a certain phenomenon to appear, even if it could be predicted quite plausibly by theoretical deduction.2 NOTES ’ The research group ZISA was supported by two research grants to the present author. In 1977-78, the crosssectional study was funded by the Minister fur Wissenschaft und Forschung des Landes Nordrhein-Westfalen; part of the work on the cross-sectional study and the longitudinal study was supported by a grant (1978-1982) from the ‘Stiftung Volkswagenwerk’. During the period when this paper was written, ZISA consisted of four researchers with whom I had the pleasure to work: Harald Clahsen, Klaus-Micheal Kopcke, Howard Nicholas and Maryse Vincent. ’ A preliminary version of this paper was presented at the European-North American Workshop on CrossLinguistic Second Language Acquisition Research, UCLA (Lake Arrowhead) 4-17 September 1981. I wish to thank Howard Nicholas, Kathleen Conners and Harald Clahsen for their comments on earlier drafts.

REFERENCES ANDERSEN, R. 1978 The relationship between first-language transfer and second-language overgeneralization-data from the English of Spanish-speaking learners. Paper presented at the Colloquium on the Acquisition and Use of Spanish and English as First and Second Languages, TESOL Convention, Mexico City, 4-9 April. ANDERSEN, R. 1979 Two perspectives on pidginization as second-language acquisition. Paper read at the Colloquium on Research on the Acquisition and Use of a Second Language Under Different Circumstances, TESOL Convention, Boston, 27 February. ANDERSEN, R. 1980 Creolization as the acquisition of a second language as a first language. In A. Valdman and A. Hightield (Eds), Theoretical Orientations in Creole Studies, pp. 273-295. Academic Press, London. ANDERSEN, R. (Ed.) Proceedings of the First European-North Second Language Acquisition Research. Newbury House, Rowley,

American

Cross Linguistic

Workshop

on

M.A.

BAUSCH, K.-R. 1971 Aspects d’une typologie de I’interference. Xeroxed Paper, Saarbrucken, Universitat des Saarlandes. BAUSCH, K.-R. 1972a Zum augenblicklichen Verhaltnis von kontrastiver Linguistik, Interferenzforschung und Fehleranalyse. Versuch eines tiberblicks. Paper read at the Workshop on Wo steht die Sprachwissenschaft?, Klagenfurt, Juni. BAUSCH, K.-R. 1972b Arbeitsbericht iiber ein Forschungsprojekt zur deutsch/franzosischen und franzosisch/ deutschen Fehleranalyse. Paper read at the Workshop on Wo steht die moderne Sprachwissenschaft?, Klagenfurt, Juni. BEVER, TH. G. 1970 The cognitive basis for linguistic structures. In J. R. Hayes (Ed), Cognition and the Development of Language pp. 279-362. John Wiley, New York. BEVER, TH. G. and TOWNSEND, D. 1979 Perceptual mechanisms and formal properties of main and subordinate clauses. In W. Cooper and E. Walker (Eds), Sentence Processing: Psycholinguistic Studies Presented to Merrill Garrett. Academic Press, New York.

45

TRANSFERAS A SECOND-LANGUAGESTRATEGY BICKERTON, Valdman (Ed), BICKERTON, Pidgin Syntax.

D. 1977 Pidginization and creolization: language acquisition and language universals. In A. Pidgin and Creole Linguistics pp. 49-69. Indiana University Press, Bloomington. D. and ODO, C. 1976 Change and Variation in Hawaiian English: General Phonology and Vol. 1 of Final Report on NSF Grant No. GS-39748.

CLAHSEN, H. 1980 Psycholinguistic aspects of L2 acquisition. In S. W. Felix (Ed), Second Language Deveiopment. Trends and Issues pp. 57-79 Narr, Ttibingen. CLAHSEN, H. In press The acquisition of German word order: a test case for cognitive approaches to L2 development. In R. W. Andersen (Ed.), Proceedings. CLAHSEN, H. 1982 Der Erwerb der Syntax in der frtlhen Kindheit. Narr, Tubingen. CLAHSEN, H., KGPCKE, K. M., MEISEL, J. M., NICHOLAS, Sprachentwicklung in der zweiten Sprache. Narr, Tubingen. CLAHSEN, H. and MEISEL, J. M. 1979 Eine psycholinguistische Papiere zur Linguistik 21,3-25. CLAHSEN, H.. MEISEL, J. M. and PIENEMANN, auslandischer Arbeiter. Narr, Tubingen.

H. and VINCENT, Rechtfertigung

M. forthcoming.

von Worstellungsregeln.

M. 1983 Deutsch als Zweitsprache. Der Spracherwerb

CONNORS, K. 1980 Native vs. non-native errors and the Ll = L2 hypothesis. Paper presented at 1979 LACUS Forum. FELIX, S. W. 1978 Linguistische Untersuchungen zum natiirlichen Zweitsprachenerwerb. Fink, Mtinchen. FELIX, S. W. 1980 Interference, interlanguage and related issues. In S. W. Felix (Ed.), Second Language Development. Trends and Issues, pp. 93-107. Narr, Tubingen. FELIX, S. W. In press. Competing cognitive structures in second language acquisition. In R. W. Andersen (Ed.), Proceedings. FITZGERALD, W. A. 1980 Nativization and second language acqusition. University, Carbondale.

M. A. Thesis, Southern Illinois

FODOR, J. A., BEVER, TH. G. and GARRETT, M. F. 1974 The Psychoiogy of Language. An Introduction to Psycholinguistics and Generative Grammar McGraw-Hill, New York. GILBERT, G. 1980 Transfer in second language acquisition. Comments on Meisel. In R. W. Andersen (Ed.), Proceedings. GILBERT, G. and ORLOVIC, M. 1975 Pidgin German spoken by foreign workers in West Germany. The definite article. Paper presented at the International Congress on Pidgins and Creoles, Honolulu, Hawaii, b-10 January. GIVEN, T. In press. Function, structure and language acquisition. In D. I. Slobin (Ed.), Cross-Linguistic Study ofLanguage Acquisition. Lawrence Erlbaum, Hillsdale, N. J. HALLE, M., BRESNAN, J. W. and MILLER, G. A. (Eds), 1978 Linguistic Theory and Psychological Reality. MIT Press, Cambridge, MA. HATCH, E. 1974 Second language learning-universals? Working Papers on Bilingualism 3, l-17. JGRDENS, P. 1977 Rules, grammatical intuitions and strategies. Interlanguage Studies Bulletin 2,5-76. JORDENS, P. 1980. Interlanguage research: interpretation or explanation. Language learning 30, 195-207. JORDENS, P. and KELLERMAN, E. 1981 Investigations into the strategy of transfer in second language acquisition. In J. J. Savard and L. Laforge (Eds), Actes du Vieme Congres International de Linguistique Appliquee, pp. 195-215. Les Presses de l’universitt Laval, Quebec. KELLERMAN, E. 1977 Towards a characterisation of the strategy of transfer in second language learning. Interlanguage Studies Bulletin 2,58-145. KELLERMAN, sition 2, 37-58.

E. 1979 Transfer and non-transfer:

where are we now? Studies in Second Language Acqui-

KELLERMAN, E. 1980 an eye for an eye. Paper presented at the Second International Colloquium ‘Acquisition d’une langue etrangere’, Paris-Vincennes, April 1980. KEUNECKE, U. 1980 Zur mbglichen Beeinflussung von Erwerbsfolgen durch den Fremdsprachenunterricht. Thesis, Universitat Wuppertal. KRASHEN, ST. D. 1977a The monitor model for adult second language performance. In M. Burt, H. Dulay and M. Finocchiaro (Eds), Viewpoints on English as a Second language, pp. 152-161. New York. KRASHEN, ST. D. 1977b Some issues relating to the monitor model. In H. D. Brown, C. Yorio and R. Crymes (Eds), On TESOL ‘77, pp. 144-158. Washington, DC. MEISEL, J. M. 1977 Linguistic simplification. A study of immigrant workers’ speech and foreigner talk. In P. S. Corder and E. Roulet (Eds), The Notions of Simplification, Interlanguages and Pidgins, and Their Relafion to Second Language Pedagogy, pp. 88-l 13. Droz, Gentve.

46

JijRGEN

M. MEISEL

MEISEL, J. M. 198Oa Strategies of second language acquisition: more than one kind of simplification. R. W. Andersen (Ed.), Pjdg~n~zat~onand Creolization as Language Acquisition. Newbury House, Rowley, MA. (Prepublication: ~u~~rtaler_Arbe~tspap~erezur S~rachwjsse~sch~i 3, I-53). MEISEL, J. M. 1980b A linguistic encounter of the third kind: will the non-real interfere with what the learner does? A Reply to Comments on Meisel. R. W. Andersen (Ed.), Pidginization and Creolization as Language Acquisition. Newbury House, Rowley, MA. MEISEL, J. M. forthcoming. Reference to past events and actions in natural second language acquisition Manuscript, University of Hamburg. MEISEL, J. M., CLAHSEN, H. and PIENEMANN, M. 1981 On determining developmental stages in natural second language acquisition. Sfudies in Second LanguageAcquisition 3 109-135. MEYER-INGWERSEN~ J., NEUMANN,~ R. and KUMMER. M. 1977 Zur sprucfhentwickfung tiirkischer Schiiier in der 3u~d~repub~ik, 2 vols. Scriptor, Kronberg/Ts. MUHLHAUSLER, P. 1980. Structural expansion and the process of creolization. In A. Valdman and A. Highfield (Eds), Theoretical Orientations in Creole Studies, pp. 19-56. Academic Press, London. NEIS, G. 1981 Der Erwerb des Franzosischen im Unterricht. Lerner~uss~rungen, Thesis, Universitat, Wuppertai.

Ein Vergieich schriftlicher

und miindljcher

PFAFF, C. 1979a Acquisition and Development of ‘Gastarbeiterdeutsch’ by Migrant Workers and Their Children in Germany. E. Traugott, R. LaBrum and S. Shepherd (Eds), Proceedings of the International Conference on Historical Linguistics, Stanford University. PFAFF, C. 1979b Constraints on language mixing: intrasententi~ code-switching and borrowing in Spanish/ English. Language!%, 291-318. PFAFF, C. and PORTZ, R. 1979 Foreign children’s acquisition of German: universals vs interference. Paper presented at the LSA Annual Meeting, Los Angeles, 27-29 December. FIENEMANN, M. 1981 Der Zweitspracherwerb ausltindischer Arbeiterkinder. Bouvier, Bonn. FGFLACK, SH. 1980 Sometimes I’ll start a sentence in Spanish y termino en Espanol. Linguistics 18, 581-618. ROHDE, U. 1980 Der naturliche Erwerb des Franzosischen als Zweitsprache bei Erwachsenen und Kindern. Thesis, Universitit, Wuppertal. SCHUMANN, J. 1978 The P~dgjn~at~on Process: A Mode~for Second Language Acqu&ition. Newbury House, Rowley, MA. SLOBIN, D. 1. 1973 Cognitive prerequisites for the development of grammar. In Ch. A. Ferguson, and D. I. Slobin (Eds.), Studies of Language Development, pp. 175-208. Holt, Rinehart, Winston, New York. SLOBIN, D. I. 1977 Language change in childhood and in history. In J. Macnamara (Ed.), Language Learning and Thought. pp. 185-214. Academic Press, New York. SLOBIN, D. 1. 1980. Universal and particular in the acquisition of language, To appear in L. Gleitman, and E. Wanner (Eds), Language Acquisition: State of the Art. Cambridge University Press, Cambridge. STEINMULLER, U. 1981 Asumans Grammatik. Zum Problem von Normverstoss und Regularitat im Zweitspracherwerb. D~k~ion Deutsch. STGLTING, W. 1980 Die Zweisprachigkeit jugosfawischer Schiiler in der Bundesrepubiik Deutschland. Harrassowitz, Wiesbaden. TAYLOR, B. P. 1975 The use of overgeneralization and transfer learning strategies by elementary and intermediate students of ESL. Language Learning 25,73-107. VINCENT, M. 1982 Les transferts: une strategic acquisitionnelle provisoire darts I’acquisition d’une langue seconde. Encrages 819,28-32. WHINNOM, K. 1971 Linguistic hybridization and the ‘special case’ of pidgins and creoles. In D. Hymes (Ed.). Pidginization and Creolization of Lunguages. pp. 91-I 1.5.Cambridge University Press, Cambridge. WODE, H. 1977 On the systematicity of Ll transfer in L2 acquisition. In C. Henning (Ed.). Proceedings of the Los AngelesSecond Language Research Forum, Dept. of English, pp. 160-169. Dept. of English, UCLA. WODE, H. 1981 Learning a Second Language. Narr., Tubingen. WODE, H. forthcoming The emergence of reference to the past in language acquisition: LI , L2, re-learning in cross-linguistic perspective. Manuscript, August 1981. ZOBL, H. 1979a Nominal and pronominal interrogation in the speech of adult francophone ESL learners: some insights into the workings of transfer. Manuscript, Moncton University. ZOBL, H. 1979b Systems of verb classification and cohesion of verb-complement relations as structural conditions on interference in a child’s L2 development. Working Papers on Biiinguafism 18.2563. ZOBL, H. 1980 The Sormai and developmental selectivity of LI influence on L2 acquisition. Lunguage Learning 30.43-57.