Explorations of Language Transfer

Terence Odlin

• Editorial: Multilingual Matters
• Año: 2022
• ISBN:
  9781788929561
• Categoría: Lingue e linguistica
• Categoría: Psicolinguistica

When learners of a new language draw on their native language (or on any other that they may know), this earlier acquired linguistic knowledge may influence their success. Such cross-linguistic influence, also known as language transfer, has long raised questions about what linguists can predict about success in the new language and about what processes are involved in using prior knowledge. This book lucidly brings together many insights on transfer: e.g. on the relation between translation and transfer, the relation between comprehension and production, and the problem of how complete any predictions of difficulty may ever be. The discussions also explore implications for future research and for classroom practice. The book will thus serve as a reliable guide for teachers, researchers, translators, interpreters, and students curious about language contact.

1 Introduction

Defining Language Transfer

Over several decades of work on second language acquisition (SLA), the term transfer has become the most commonly used shorthand to denote a phenomenon also known as crosslinguistic influence. Later in this chapter (and in more detail in the next), the history of the term itself will be considered, but for now it will suffice to say that despite controversies about the term and about crosslinguistic influence, many SLA researchers continue using the word transfer or the phrase language transfer.

Specialists in other fields have also employed transfer in a variety of ways, and, in psychology, different theorists have found different uses for the term, as a comparison of its meanings in psychoanalysis and behaviorism will show. Similarly, the uses of transfer in SLA and related fields are not at all the same. Some relevant domains will be compared, but central to the discussion of the term in SLA will be notions addressed in the working definition given here:

Transfer is the influence resulting from the similarities and differences between the target language and any other language that has been previously (and perhaps imperfectly) acquired. (Odlin, 1989: 27)

The definition deliberately uses ‘any other language’ instead of ‘a second language’ since there are many cases of people learning not only a second language but also a third. Two cases of multilingualism are considered in this book: one where English is the target language and Finnish and Swedish the previously learned languages (Chapters 5 and 7), and another where there is again one native language, Polish, and two non-native languages – German and English (Chapter 7). Along with these empirical investigations, other topics involving multilingualism are also considered in various chapters.

Multilingual research has offered new perspectives on the complexity of crosslinguistic influence (e.g. Cenoz et al., 2001; De Angelis & Dewaele, 2011; Gabryś-Barker, 2012; Hammarberg, 2009). There has also been considerable work on reverse transfer in bilingual settings. For example, Porte (2003) analyzed the English spoken by several native speakers of English who were teaching EFL (English as a Foreign Language) in Spain and encountered many cases of L2 → L1 influence: e.g. I was really shocked when I first saw how molested some teachers got at my criticising the system (Porte, 2003: 112), where molested reflects just the mildly unfavorable meaning of Spanish molestar (‘annoy’) in contrast to the highly pejorative English molest, which is often used to denote criminal behavior (e.g. child molester). The teachers in Porte’s investigation seem to have been influenced both from some direct knowledge of Spanish and from their relatively long residence in Spain, especially since the molested example might just as easily come from a native speaker of Spanish using English (Nash, 1979).

Along with L2 → L1 influences on vocabulary, interesting cases of grammatical transfer are also evident in the literature, as in a chapter by Pavlenko (2003). She found that Russian speakers living in the United Sates sometimes used the perfective/imperfective system of verbs in Russian in ways quite different from monolingual speakers in Russia, with the Russian speakers in America sometimes resembling learners of L2 Russian who were L1 speakers of English. Research on L2 → L1 transfer often overlaps with what is called code-switching, and it has proven difficult in some cases to decide if a particular bilingual utterance should be considered transfer or switching (e.g. Isurin et al., 2009). Chapter 4 of the present volume compares L2 → L1 and L1 → L2 influences involving word order.

Patterns of L2 → L1 and L1 → L2 transfer raise many interesting questions about how much parallelism in crosslinguistic influence there may be in any language-contact setting (Thomason & Kaufman, 1988). Still other forms of parallel influences can be found by comparing acquisition in different settings – for example, Italian children who study English in England, and English-speaking children who study Italian in Italy. Rocca (2007) looked closely at just such cases, with regard to tense and aspect structures in the target language, and she identified clear patterns of transfer.

What researchers regard as typical cases of transfer – and what probably concern language teachers the most – are found in divergences between the target language (e.g. English in EFL settings) and the source language (the native language in L2 acquisition, and either the L1 or L2 in L3 acquisition). Transfer errors in either comprehension or production are often called negative transfer. Such errors frequently appear in vocabulary uses, such as when a native speaker of Spanish chooses molest in English as a synonym of annoy. Along with negative transfer involving vocabulary, divergences can and do appear in grammar, as in the following sentence from a native speaker of Vietnamese: She has managed to rise the kite fly over the tallest building (= She has managed to fly a kite over the tallest building) where the pattern rise… fly strongly suggests influence from a key pattern Vietnamese syntax known as serial verb constructions (Helms-Park, 2003). The pronunciation and spelling of learners also indicate many cases of negative transfer. Spelling errors frequently offer a window on pronunciation problems, as in an error of a Finnish student who wrote crass instead of grass. Thus a Finn’s misspelling of grass with either the letter <c> or the letter <k> reflects a phonological difference between Finnish and English: the former does not make a phonemic distinction between /k/ and /g/.

Although negative transfer often gets more attention from researchers and teachers, there is also the reality of positive transfer, i.e. a convergence between what speakers of the target language normally do and what non-native speakers succeed in doing because of some similarity between the source language and the target. Examples of positive transfer will appear further on in the discussion of predictions.

The Transfer Metaphor

Although the term transfer has become the most commonly used designation for crosslinguistic influence, it has been controversial, as reflected in criticism by Corder, who asserted that ‘nothing is being transferred from anywhere to anywhere’ (1983: 92). Indeed, transfer is a metaphor when applied to crosslinguistic influence and to other phenomena in linguistics and psychology; in the latter field, for example, researchers have investigated the use of old skills for solving new problems (Singley & Anderson, 1989).

The metaphoric basis of transfer comes into sharper focus through a look at its etymology. Transfer reflects two Latin forms, trans (across) and ferre (carry), so that there is a suggestion of something being carried from one place to another. In fact, the English phrase carried across has served as an alternative expression for crosslinguistic influence in some discussions of SLA (e.g. Slobin, 1996), and at least one French linguist (Meillet, 1933) used the verb transporter in exactly the same way.

The notion of carrying across has also had other extensions, as seen in the very word metaphor. Dechert (2006) observes that the etymologies of both transfer and metaphor literally denote carrying over, transfer coming from the Latin forms given above, and metaphor from the Greek meta (across) and phor (carry). The notion of movement implicit in both words makes them (along with carry over and transport) part of a large set of expressions that Reddy (1979) called ‘the conduit metaphor,’ examples of which also include expressions such as get something across as a synonym for explain. The motion implied in transfer can also help understand the metaphoric basis of the verb translate, which has essentially the same etymology as transfer, since the word latus is a participial form of the Latin irregular verb ferre. Despite the objections of Corder and others, the transfer metaphor as well as the translation metaphor is useful. As Lakoff and Johnson (1980) argue more generally, metaphors offer a tool to try to understand new problems with the help of existing cognitive resources. The phenomena for which transfer and translate serve as cover terms are complex, but both metaphors rightly hint at interactions between languages or, more precisely, interactions in the minds of bilinguals or multilinguals.

The conduit metaphor expressed by transfer and translate also has interesting counterparts in German. The word übertragen (über ‘over’ + tragen ‘carry’) is sometimes equivalent to translate and sometimes to transfer in the sense of crosslinguistic influence. Furthermore, a related word – hinübertragen – has been used to denote both carrying and crosslinguistic influence. It appears in what may be the earliest discussion of language transfer as a psycholinguistic phenomenon:

Die erlernung einer fremden Sprache sollte daher die Gewinnung eines neuen Standpunktes in der bisherigen Weltansicht sein, und ist es in der That bis auf einen gewissen Grad, da jede Sprache das ganze Gewebe der Begriffe und die Vorstellungsweise eines Theils der Menschheit enthält. Nur weil man in eine fremde Sprache immer, mehr oder weniger, seine eigne Welt-, ja seine eigne Sprachansicht hinüberträgt, so wird dieser Erfolg nicht rein und vollständig empfunden. (von Humboldt, 1836: 59; emphasis added)

[The learning of a foreign language should thus be the attainment of a new perspective on the worldview already held and indeed it is to a certain degree, since every language contains the entire fabric of concepts and forms of representation of a part of humanity. Still, because one always transfers, more or less, one’s own worldview and, surely, one’s own language-view, this success will not be felt pure and complete.]

This passage, taken from a well-known treatise on language and mind by the philosopher Wilhelm von Humboldt, not only uses hinübertragen to refer to transfer but also raises issues that have eventually come to be examined rather closely in modern SLA research, issues sometimes referred to as conceptual transfer (e.g. Jarvis & Pavlenko, 2008; Lucy, 2016; Odlin, 2005). While Humboldt used hinübertragen, other intellectuals in the 19th century employed übertragen to denote transfer or translation (or both), including Hugo Schuchardt and Hermann Paul, as will be seen in Chapter 2. The same chapter provides detailed evidence that the use of transfer in English in discussions of language contact and SLA owes more to the German use of hinübertragen and übertragen than to the behaviorist theories of learning in psychology that are often cited in the history of transfer research.

The coincidence in German of one word – übertragen – denoting both transfer and translation should not, of course, be taken as proof that transferring and translating are really the same thing. Even so, it would also be mistaken to assume that the two phenomena are completely different. Chapter 7 offers what may be the first in-depth look at the overlap between transfer and translation.

Predictions of Transfer

One of the most controversial questions regarding transfer has been whether it is possible to predict when crosslinguistic influence will occur. The chapters in Part 1 consider different aspects of the issue. Chapter 2 looks at predictions through a historical lens, focusing on the so-called Contrastive Analysis Hypothesis, a term which the chapter shows to be quite nebulous. Chapter 3 considers whether a contrastive analysis (i.e. the systematic comparison of two or more languages) could ever be complete, and it also argues that putative constraints on transfer are themselves predictions. For example, one supposed constraint claims that there can be no transfer of basic word-order patterns. Chapter 4, however, offers evidence to the contrary, thus calling into question the prediction of no transfer.

Although the putative constraint on the transfer of word order is contradicted by actual evidence of such transfer (and likewise other supposed constraints discussed in Chapter 3), there is no reason in principle for denying the possibility of some constraints on crosslinguistic influence. By the same token, however, there is also no reason to believe that predictions of transfer are always doomed to failure. Here is one prediction made in a recent discussion (Odlin, 2014: 30):

• Speakers of Finnish as a group will have greater difficulty with the articles of Portuguese than will speakers of Swedish as a group.

The prediction seems plausible, given the existing evidence for a difference in success between Swedish and Finnish speakers in the acquisition of English articles (Chapter 5). The Portuguese article system is, of course, different in some ways but, like English, it does have obligatory articles, some of which signal definite and others indefinite reference.

The prediction considered here, as well as additional ones in Chapter 3, should be understood as expectations of group, not individual, tendencies. That is, positive transfer is predicted for one group whose language shares a structural trait with the target language. Like English, Swedish has an article system whereas Finish does not, and so Swedish speakers should, if the prediction holds true, tend to do better.

Although shared traits such as articles do not guarantee transfer, the comparative group approach suggests that the method can often be used to verify or falsify cases of crosslinguistic influence. A study by Jarvis (2002) of article use shows a further methodological refinement: two L1 groups, Finnish and Swedish, are compared in how they differ in their L1 behaviors as well as in how they differ in their performance in the English target language. By comparing reference strategies used by native speakers of Finnish and Swedish in their own L1s, Jarvis was able to correlate these strategies with the ways that Finnish and Swedish speakers used – or did not use – articles when writing in English. Much of Chapter 5 and also parts of Chapters 6 and 7 of this volume consider cases of Finnish and Swedish influence to be found in the database compiled by Jarvis.

Although controlled comparisons of L1 groups offer an especially useful way to study transfer, other types of investigations can also provide important insights, such as studies of language-contact situations. One example is the so-called after-perfect, where after marks perfective aspect as in this sentence from a speaker in the Western Isles of Scotland: The stone is after going through, he says (Sabban, 1982: 155), which could be paraphrased as The stone has gone through…. Similar after-perfect sentences are also common in Ireland, and there is a close parallel between the Celtic English structure and patterns in both Irish and Scottish Gaelic. In a survey of worldwide varieties of English, Kortmann and Szmrecsanyi (2004: 1151) found that the after-perfect appeared only in the Celtic lands or in areas where speakers of Irish and Scottish English lived, especially certain parts of Canada. In contrast, the same global survey found some verb forms to be extremely widespread, such as verbs that are traditionally irregular being made into regular forms, as where the past tense of catch becomes catched instead of caught; this pattern is common among both native- and non-native speakers of English. Although it would be hard to argue for transfer in the case of catched, the after-perfect does indicate contact-induced change due to transfer.

In recent decades the academic fields often labeled as language contact and second language acquisition have drifted apart, even though some researchers remain active in both fields (e.g. Mufwene, 2010). However, all cases of SLA involve some kind of language contact, and pioneers in the study of transfer such as Hugo Schuchardt and Uriel Weinreich (Chapter 2) recognized the value of both classroom-based and naturalistic research. The wide domain of language contact seems all the more important to acknowledge if researchers wish to get a good grasp of the problem of predictions. Because the various proposed constraints on transfer have generally come from research done in school settings, work in naturalistic settings can bring to bear a wider range of data from second language users, as in a number of studies of word-order transfer cited in Chapter 4. Many regions of the world offer potential opportunities to study contact-induced variation and change due to transfer, but only a small number have received much attention, such as the Celtic lands of the British Isles and the Andes region of South America (e.g. Filppula et al., 2008; Klee & Ocampo, 1995; Muysken, 1984). There seems little doubt that if research can be carried out in more diverse settings, a wider range of findings about transfer will emerge.

Transfer and Language Processing

Issues related to predictions inform the chapters of Part 1, while in in Part 2 issues of language processing are prominent, specifically, comprehension and production (Chapter 5), using and understanding focus constructions (Chapter 6), and translation (Chapter 7). Transfer researchers have long acknowledged the importance of processing. Weinreich (1953a) and S...