Definitions

Along with the particular studies of transfer cited so far, there have appeared review articles of transfer (Odlin, 2001, 2002, 2003) focusing on research that appeared since earlier book-length treatments of crosslinguistic influence (Odlin, 1989; Ringbom, 1987). All of these studies discuss problems of defining transfer and related terms, and interested readers may consult those discussions. In this chapter, however, just three terms will be considered: linguistic relativity, conceptual transfer, and meaning transfer.

Linguistic relativity is often defined as the hypothesized influence of language on thought. Such influence might affect either comprehension or production, and such influence could, of course, affect comprehension or production in a second language (or a third, a fourth, etc.); moreover, the influence might be where the L1 is influenced by the L2. Conceptual transfer can accordingly be defined as those cases of linguistic relativity involving, most typically, a second language.

Caution is necessary in analyzing the connection between transfer and relativity. Cases of influence from the semantics or pragmatics of the native language (or a second in L3 acquisition) constitute what can be called meaning transfer (a term neutral between semantic and pragmatic influence). However, not all cases of meaning transfer involve linguistic relativity. One clear instance showing the need to distinguish meaning transfer and conceptual transfer is seen in an interlanguage English pronoun in Poland. Polish uses a first-person plural ending (-my) in some cases with a singular referent as in the following example:

The byliśmy suggests more than one referent, but speakers of Polish often construe the meaning as a singular in some discourse contexts. Thus a possible transfer error would be We were at the theater with my brother yesterday when the first-person referent is singular. In fact, the Polish sentence (Wczoraj byliśmy) is discussed in a Polish textbook that goes on to make the following observation in ellipsis:

The “carrying over” clearly involves semantic as well as pragmatic transfer, but it would be absurd to claim that Poles do not conceptually distinguish singular and plural referents.

The Polish example thus warrants the following generalization: All conceptual transfer involves meaning transfer but not all meaning transfer involves conceptual transfer. In effect, conceptual transfer is a subset of meaning transfer. The generalization squares with other arguments such as one of Levinson (1997) that “semantic representations” must be theoretically distinct from “conceptual representations,” and the same likely holds true of pragmatic representations, although Levinson does not address that issue. If meaning and thought were theoretically equivalent notions, any relativistic claim about language influencing thought would be tautological. Much of this chapter will examine types of meaning transfer that may or may not also be cases of conceptual transfer. Before examining such cases, however, a look at the history of relativity in relation to transfer will provide some useful background.

Von Humboldt, Whorf, and Subsequent Empirical Work

In the study of linguistic relativity the two best-known theorists are the German Wilhelm von Humboldt (1767–1835) and the American Benjamin Lee Whorf (1897–1941). Both von Humboldt and Whorf developed their interest in relativity largely as a result of their study of other languages, and in fact both regarded conceptual transfer as one manifestation of relativity even though their views on transfer differed somewhat.

Von Humboldt's most detailed discussion of relativity is a book-length preface (1836/1960, 1836/1988) that he wrote for his three-volume linguistic and cultural study of Kawi, a literary language in Java (von Humboldt, 1836–1839). Although obviously fascinated with the East Indies (which he never actually visited), von Humboldt has little specific to say about Kawi in the preface and instead foregrounds the diversity of human languages and the implications he sees in that diversity for the relation between mind and language. In the ninth chapter of the

preface, he views the learning of a foreign language as the attainment of a new perspective in one's existing world-view (Weltansicht, p. 75). In fact he sees second language acquisition as the only way an individual may escape from the conceptual world of the native language. Yet the escape is never, in the view of von Humboldt, completely successful: because “one always more or less carries over (hinüberträgt) one's own world-, indeed one's own language-view (Welt-…Sprachansicht), this success will not be felt [to be] pure and complete” (1836/1960, p. 75, my translation). The verb hinüberträgt is an early use of the metaphor of transfer (the Latin source of which likewise means carrying over), and the use of Welt-…Sprachansicht shows that von Humboldt considered the transfer to be both conceptual and linguistic (cf. Dechert, 2004).

Whorf's views resemble von Humboldt's in some respects, but not in all. One characteristic of the mind/language relation that Whorf emphasizes is how most linguistic knowledge is unconscious. In one article, (Whorf, 1940/1956) he notes the difficulties of a typical English-speaking learner of French who is unaware of native language patterns which continue to inhibit learning. Whorf then optimistically says of the learner,

Unlike von Humboldt, Whorf believed that consciousness raising can overcome inhibiting influences of the native language. However, Whorf took seriously the likelihood that the “binding power” of the native language could also affect the metalanguage used to characterize the structures of a language being studied. In another article Whorf writes: “We tend to think in our own language in order to examine the exotic language” (1941/1956, p. 138). His own experience with Hopi shows how he became aware of the analytical dangers of metalinguistic transfer (Odlin, 2004a). In the mid-1930s he analyzed Hopi as a language with three tenses: “factual or present-past, future, and generalized or usitative” (Whorf, 1936/1956, p. 51). However, his reanalysis of about two years later held that the verb system was not based in tense but in categories of evidentiality (“reportive,” “expective,” and “nomic”), which are, as he put it, “distinct realms of validity” (Whorf, 1938/1956, p. 113) that are “mutually exclusive” (p. 115). The experience of revising his analysis made him aware of the danger of spurious interlingual identifications: “Hopi categories are just enough like Indo-European ones to give at first a deceptive impression of identity” (p. 112).

For Whorf as for von Humboldt, the influence one's language extends beyond metalanguage:

His notion of “habitual” thought might sound behaviorist, but elsewhere he made his skepticism about behaviorism clear, and indeed his ideas seem more consonant with recent ones framed within the theoretical concept of “automaticity” (e.g., Dewaele, 2001; Hammarberg, 2001; Segalowitz, 2003).

Whorf's ideas and those of other relativists have generated many controversies. Some doubts about relativity arise from empirical concerns, as in the detailed analysis of Hopi by Malotki (1983) which calls into question Whorf's later belief that the Hopi verb is “timeless.” Another skeptic (Pinker, 1994) cites a range of psycholinguistic studies that he sees as counterevidence to relativistic claims. Other objections go beyond empirical issues; as Lakoff (1987) notes, “conceptual relativism is often confused with moral relativism, and therefore any relativism is seen as the denial of the possibility of universal standards of ethical conduct” (p. 326). Certain philosophers and literary theorists (e.g., Battersby, 1992; Davidson, 1984) have pointed to shortcomings in some relativistic approaches in their fields. For example, one literary determinist (Fish, 1985) would have it “that the interpreter sees and hears within and as an extension of those [socially constructed] interpretive systems—he cannot choose for or against them because they constitute the limits of his consciousness” (p. 125, emphasis in the original). Fish assumes that individuals can belong to one and only one interpretive community, but the existence of bilingual and multilingual communities shows such an assumption to be wrong. Moreover, acquiring a second language can affect one's native language, as a great deal of research from Weinreich (1953/1968) to Cook (2003) has shown.

Despite the problems noted by skeptics such as Pinker and despite the dogmatism of determinists such as Fish, empirical work on relativism has made considerable progress in the last two decades. Work by John Lucy (1992) offered convincing evidence of certain differences between speakers of English and Yucatec with regard to how people categorize and remember certain kinds of objects, and how characteristics of the grammar of noun phrases in each language affect those differences in cognition, and recent work by Lucy and Gaskins (2003) on Japanese and English offers further support for the Whorfian notion of habitual thought. In a similar vein, research by Pederson et al. (1998) found that speakers of typologically different languages performed quite differently in a communication task designed to explore how speakers interpret spatial arrays, and the performances covaried with the type of locational system in the native language. Thus whereas speakers of Dutch

used their deictically based system of left and right to recall the location of objects, speakers of Arandic (an Australian language) used their language's system of absolute locations (comparable to the cardinal points of north, south, etc.) to recall the same objects. Speakers of languages relying on absolute systems such as Arandic were found to have more accurate recall of the location of objects in comparison with speakers of languages such as Dutch. One of the members of the Pederson research team, Stephen Levinson, has organized similar research projects which will be discussed in the next section.

In second language acquisition there has long been an awareness of the possible significance of relativism, as seen in discussions in the mid-20th century by Lado (1957) and Kaplan (1966/1984), as well as in very recent discussions (e.g., Kramsch, 2004). Over the years there have been various empirical attempts to demonstrate the influence of L1 concepts, as in a study of putative effects of grammatical gender on the acquisition of English (Clark, Losoff, McCracken, & Still, 1981). One study, especially interesting because of its methodology (Ijaz, 1986), considered cases where meaning transfer from two native languages, Urdu and German, affected the use of spatial prepositions in English; in some cases, learners agreed in their prepositional choices but in others, diverging choices by the two groups reflected differences between the native languages. Ijaz assumed that meaning transfer was the same as conceptual transfer, and as the discussion in the preceding section suggests, such an assumption is mistaken. Furthermore, the cloze tests Ijaz used cannot really show the influence of a linguistic system on nonverbal cognition. Recent research discussed in the next section avoids this methodological problem by using elicitation devices such as picture description.

In the 1990s, the interest in conceptual transfer intensified. Slobin (1993) emphasized the fact that different languages make certain kinds of meaning more salient than others, and he viewed first language acquisition as a process of children developing varied routines of “thinking for speaking;” the outcome of the process is, in Slobin's view, an L1-specific world view likely to affect the subsequent acquisition of another language. The similarity of Slobin's approach to von Humboldt's is not accidental: He cites the German philosopher on this very point. Kellerman (1995) coined the phrase “transfer to nowhere” for cases of learners attempting to use an L1 meaning category that is not highly congruent with an L2 conceptual category (cf. Odlin, 2003). Although he did not invoke Whorf, Kellerman also stressed the unconscious nature of L1 categories involved in “transfer to nowhere.”

Relativistic ideas also informed the empirical studies of Jarvis (1998) and Pavlenko (1999). In a comparison of transfer patterns in the English of L1 Finnish and L1 Swedish speakers, Jarvis found the two groups often differing in verb and prepositional choices to describe motion events and spatial relations. Jarvis interpreted such differences as rooted in L1-based conceptual differences, with the native language inducing different patterns of cognizing the events that learners described. Pavlenko likewise attributed to L1-based manners of cognizing the differences she observed in the descriptions of events by speakers of Russian and

English. There will be further discussion of these as well other studies in the next section.

Recent Work

Not all meaning transfer should be seen as conceptual transfer, as the Polish English example discussed earlier shows. Even so, some cases of semantic or pragmatic transfer seem to overlap with conceptual transfer, even if it proves difficult to determine the overlap in some particular cases. Three areas of meaning transfer have gotten considerable recent attention: meanings related to space, time, and affect. In each of the three following subsections, issues of transfer will be discussed in relation to possible universals as well to relativity. Although relativism and universalism are sometimes viewed as incompatible theories, the soundest approaches to the study of human diversity have never rejected the notion of a fundamental unity of the human species (and in fact neither von Humboldt nor Whorf rejected such a unity). By the same token, the study of conceptual transfer and universals should be seen as interdependent (Odlin, 2002).

Spatial Meanings

Until recently there were few challenges to the belief that human beings represent spatial meanings in the same way even though the grammatical realizations vary considerably across languages. One fairly recent universalist stance is evident in a paper of Clark (1973), who developed a neo-Kantian phenomenology of space based on the canonical orientation of the upright human, and several other linguists have relied on this anthropomorphic approach (e.g., Levelt, 1989, Lyons, 1977). Even Whorf conceded some universality to the experience of space: “probably the apprehension of space is given substantially in the same form by experience irrespective of language” (1941/1956, p. 158). Nevertheless, he distinguished between the “apprehension” (i.e., perception) of space and conceptions of space, which he did view as somewhat language-specific.

Much of the recent relativistic work has looked in detail at how nonlinguistic spatial cognition covaries with crosslinguistic differences in linguistic structure. For example, Levinson, Meira, and The Language and Cognition Group (2003) found little crosslinguistic uniformity in the semantics of the adpositions used by speakers of nine typologically distinct languages, all of whom had been given a picture description task designed to be cross-culturally neutral. In a book-length study, Levinson (2003) expands on the findings of Pederson et al. (1998), reporting follow-up work with larger groups of individuals, with a larger sample of languages, with more kinds of experimental tasks, and with procedures to distinguish possible environmental influences only indirectly related to language (e.g., type of schooling). The results proved consistent with those of the Pederson et al. study. Speakers of languages such as the Austronesian language Longgu and the Mayan language Tzeltal performed very similarly on tasks which tested recall and recognition memory of the spatial arrangement of objects, both groups preferring a locational strategy of absolute coordinates (comparable to north, south, etc.). In contrast,

speakers of both Dutch and Japanese preferred a relative locational strategy (based on right/left deictic orientations).

The Levinson team has confined itself largely to investigations of monolinguals' conceptions of space. However, other work has considered what occurs with people who have acquired or are acquiring a second language. Carroll, Murcia-Serra, Watorek, and Bendiscioli (2000) have inferred differences in spatial cognition in native speakers of German, on the one hand, and in English and Spanish speakers and their L2 German on the other, this inference coming from the results of picture-description tasks. For example, although the L2 users' morphosyntactic production of most structures in German was formally accurate, they produced (in contrast with a group of native speakers) few coadverbials such as daneben (beside there) as in Auf dem Platz ist ein Kiosk; daneben is[t] ein Farbandständer (On the square is a kiosk; beside there is a bicycle stand). The da in daneben is historically cognate with English there, and in fact some German coadverbials such as dazu and darin have Germanic cognates in English as in thereto and therein; even so, the German system plays a much larger role in native speakers' linguistic representation of space. The less productive English coadverbials may offer some positive transfer: on some spatial tasks English speakers approximated the German norms more closely than did a group of speakers of Spanish, a language considered by Carroll et al. to be like English its spatial system. Nevertheless, the English as well as the Spanish groups' underproduction of coadverbials supports von Humboldt's surmise about how difficult it is to fully adopt a new Welt-…Sprachansicht (world and language view).

The picture description task of Carroll et al. resembles that of Levinson et al. (2003) and offers a more promising approach to study conceptual transfer than does the kind of test used by Ijaz, described earlier. A film used by Jarvis (1997, 1998) and other researchers provides similar methodological advantages. Like the Carroll team, Jarvis studied native as well as nonnative spatial representations, and he identified some consistent differences between L1 Finnish and L1 Swedish speakers in their English narrations of events in a Charlie Chaplin film. For instance, Jarvis found that the L1 Swedish group often chose the prepositional verb run on whereas the L1 Finnish group did so much less often (and all those who did were learners who had also studied Swedish). The use or nonuse of run on concurs with the lexical preferences in L1 narrations of the film in Swedish and Finnish. Furthermore, Jarvis and Odlin (2000) found a comparable difference in some prepositional choices in the same narratives: in a scene showing Chaplin and Paulette Goddard on a lawn, the phrase sat to the grass was chosen by several Finnish speakers but not by any Swedish speakers.

All the second language research discussed so far has focused on linguistic representations of space, but paralanguage is another option that speakers can and do use. In a study of gestures used with speech, Kellerman and van Hoof (2003) identify what seem to be clear cases of transfer of L1 Spanish gestures (as seen among native speakers who were videotaped) in speakers' use of L2 English spatial descriptions of events in a story presented in pictures. Not all the results that

Kellerman and van Hoof obtained were so straightforward, however. As the authors observe, there was little if any L1 transfer in other groups studied (L1 English speakers using Dutch, and L1 Dutch speakers using English). Kellerman and van Hoof candidly admit that they cannot explain the inconsistency of their results. One problem may be in their typological assumptions, which are also those of Slobin, who used a typological framework developed by Talmy (1985). In that framework, languages such as Dutch and English resemble each other more closely than does Spanish. Readers may recall, however, that the research of Carroll et al. (2000) posited a greater typological similarity in spatial representation between English and Spanish than between English and German (the latter being a language not very different from Dutch). Despite the discrepancies (which obviously call for a closer look at the typologies), some results of the gesture study are consistent with relativistic claims of Kellerman (1995, 2001), and will likely lead to further work.

Temporal Meanings

There have been several attempts to construct theoretical frameworks for the meanings of tense and aspect (e.g., Chung & Timberlake, 1985; Comrie, 1985, 1976), but the ideas having the greatest impact on second language acquisition are formulated in the Aspect Hypothesis (Andersen & Shirai 1996; Bardovi-Harlig, 2000). Owing much to analyses of Vendler (1967) and Dowty (1979), the Aspect Hypothesis maintains that the evolution of learners' use of tense and aspect reflects their associations between the forms marking tense/aspect and the inherent lexical aspect of the verbs that they use. For example, the hypothesis predicts that learners of English will tend to mark atelic (i.e., unbounded) meanings with progressive forms (e.g., We're working on the project) and to mark punctual meanings with past forms (e.g., We finished the project yesterday). According to the hypothesis, cognitive simplicity and prototypicality underlie such associations.

Empirical work on the Aspect Hypothesis has shown an impressive if not total consistency in studies of learners of many different language backgrounds. Proponents of the hypothesis have not usually denied the possibility of transfer involving tense and aspect, but as Shirai and Nishi (2003) observe, the emphasis of such research has been on universality, and “not much attention has been paid to the cross-linguistic differences in how these aspectual notions are lexicalised across languages” (p. 267). Their own paper is not an empirical study so much as a contrastive analysis of Japanese and English, which they naturally see as a preliminary for a study of actual transfer. As Shirai and Nishi suggest, there are real differences in how languages code temporal meanings. In fact, there are also some convincing studies of transfer effects involving such differences.

Some of the clearest evidence comes from a study by Collins (2002). As mentioned earlier, Collins studied only one native language group (French speakers), but she used the same cloze tests of English verb forms that Bardovi-Harlig and Reynolds (1995) had employed to investigate the use of the English tense and aspect by speakers of other languages. Most significantly, Collins found that the use of the perfect by French speakers was very different from that of the other groups. Unlike

those groups, the Francophone learners often chose a perfect as an alternative to past simple marking of telic verbs, a result which Collins attributes to the passé composé of French. Even so, she interprets her results as support for the Aspect Hypothesis because the perfect was not common except among learners who had already started using past forms to mark telic verbs.

Rocca (2003) likewise finds support for the Aspect Hypothesis in her longitudinal study of bidirectional transfer. Looking at the acquisition of L2 Italian by English-speaking children and of L2 English by Italian-speaking children, she stresses, like Collins, the criterion of developmental readiness: Transfer of tense or aspect does not arise until learners show considerable progress in making the canonical associations predicted by the Aspect Hypothesis (e.g., progressive forms with atelic verbs). Once they have made such associations, however, learners then may tailor the Aspect Hypothesis to correspondences between the L1 and L2 where, for example, Italian children appear to seek a correspondence between a past imperfective form such as voleva (was wanting) and a one-word progressive such as wanting as in Bunny wanting to catch the little devil.

Both the Collins and Rocca studies emphasize the role of crucial similarity as expounded by earlier researchers such as Wode (1983), who recognized the importance of developmental readiness as a prerequisite for transfer. Yet also significant is the likelihood that interlanguage forms often encode meanings of the L1 more than of the L2. In the Bunny wanting sentence, the verb may indicate past tense as well as imperfective aspect in the child's own interlanguage semantics because one-word verb forms of Italian such as voleva code tense as well as aspect. Wenzell (1989) came to a similar conclusion about the alternation of past and nonpast forms in the interlanguage of Russian learners of English who seemed to be making interlingual identifications of past forms in English with perfective ones in Russian and non-past forms with imperfective ones. Something analogous also appears in the results obtained by Klee and Ocampo (1995) in a study of Spanish verb forms used by speakers of Quechua. Although present and past perfect forms appeared frequently in the Spanish of these bilinguals, the forms often did not mark tense but rather the modal category of evidentiality.

The studies of temporal meanings discussed up to this point do not offer clear evidence of conceptual transfer even though they do indicate some meaning transfer. However, von Stutterheim (2003) sees conceptual transfer at work in a study using film narration as the elicitation device. Native speakers of German tend to frame the events in the film with end points that give closure to the events whereas native speakers of English tend not to use such endpoints. For example, native speakers of English often said Two nuns are walking down a road in contrast to German speakers, who often gave an end point as in Zwei Nonnen laufen auf einem Feldweg Richtung eines Hauses (Two nuns walk along a lane towards a house), where the house is the endpoint of the event frame. The L1 English pattern of framing seems to be transferable. English speakers often did not state an endpoint in their German sentences as in Zwei Nonnen laufen auf der Strasse lang (Two nuns walk the street along), where the house is not mentioned. The framing difference is

linked, according to von Stutterheim, to the grammatical importance of progressive aspect in English; in contrast, progressive aspect in German is not prominent. She argues that the typological difference has cognitive implications, where the prominence of progressive aspect in English induces speakers to frame events more analytically, whereas German speakers represent such events more holistically. If von Stutterheim is correct, the English progressive plays a role in such transfer but only an indirect one. She stresses that such crosslinguistic influence does not involve grammatical or semantic errors but rather differences in cognizing that persist even among very advanced learners.

Affective Meanings

The terms affective, emotional, and attitudinal are roughly synonymous, although the third term often indicates mental states that do not always involve emotion, including, for example, epistemic and deontic modality. In any case, affect permeates much of what speakers consider to be “meaning”—itself a highly ambiguous notion, as semanticists such as Lyons (1977) have often stressed, and the importance of affective meanings (as opposed to referential meanings) has been foregrounded by Ogden and Richards (1923) among others. Even though affect is not always deemed to be a cognitive phenomenon, several researchers (e.g., Lazarus, 1991; Ortony, Clore, & Collins, 1988; Schumann, 1997) see it so.

Affect has gotten recent attention from linguists interested in conceptual transfer (e.g., Pavlenko, 2002). Less studied has been the possible universality in human language of some emotions. Darwin (1872/1998) was intrigued by the idea of pan-cultural patterns of emotional expression shared to some extent with other species, and in more recent times a lively debate has arisen about universals and relativity in affect (e.g., Ekman, Levenson, & Friesen, 1983; Lutz & White, 1986; Wierzbicka, 1999). Some empirical work has attempted to detail culturally-specific readings of emotional states (e.g., Matsumoto et al., 2002) whereas other work has focused on possible pan-cultural affective associations (e.g., Moore, Ranney, Hsia, & Rusch, 1999). The anthropological study of the Moore team used word association tasks to develop statistical models of semantic space for emotion words of speakers of Chinese, English, and Japanese. They found similar lexical distances among all three groups. For example, the English words tired, bored, and lonely were adjacent to one another in contrast to the words love, happy, and excitement, which were in term clustered rather tightly. Similar results obtained for the semantic spaces of corresponding words in Chinese and Japanese. Despite the similar patterns of clustering, there were language-specific differences, as Moore et al. note. Moreover, the results from bilinguals in their study indicated that the L1-specific semantic space influenced L2 judgments, and a dissertation by Rusch (1996) that focused on Japanese-English bilinguals showed similar evidence of transfer. Neither investigation looked much at all at the literature on bilingualism or SLA, and both studies seem naïve in certain linguistic assumptions. Nevertheless, their results constitute a serious look at a likely pan-cultural basis for the interlingual identifications that learners can sometimes make in the emotion vocabulary of an L1 and L2.

Whatever universality there may be, it does not make the people's emotional lives identical. Speech communities vary in how they code the affective meanings that permeate languages, and the linguistic differences can induce conceptual as well as meaning transfer. Pavlenko (1999, 2002) has detailed a variety of forms of L1 Russian influence on L2 English and L2 English influence on L1 Russian involving affect. For example, some Russian speakers use idioms based in L1 to describe personal feelings as in she is deep inside herself, which was said of a woman in a film who appeared to be deep in thought (with the English words clearly a calque on the Russian byt' w sebe (to be inside oneself). In contrast to monolingual Russians, those who were bilingual talked about emotions differently in their L1 where, for instance, the former often used the verb stanovitsa (become, get) with adjectives of emotion such as serieznoe (serious) whereas monolinguals used emotive verbs instead of verb + emotive adjective collocations like those of English.

Although such cases might be interpreted as meaning transfer yet not conceptual transfer, other findings in Pavlenko's research make a compelling case for a relativistic interpretation. In comparison with monolingual English speakers, Russian monolinguals more frequently referred to body parts (face, hands, etc.) to assess the emotional states of characters, and bilinguals using their L2 English and L1 Russian likewise frequently referred to body parts. Moreover, Pavlenko found the Russian of bilinguals who had lived in the United States to be different from that of monolinguals in that the former but not the latter conceived of some events in a film as an invasion of personal space or of privacy, an affective concept that Pavlenko views as having no close translation equivalent in Russian. Work on compliment behavior by Yu (2004) shows a similar pragmatnic influence of the L2 on L1.

The problem of translating affect arises in grammar as well as in lexis, and focus constructions offer many examples of the elusiveness of cross-linguistic equivalence. Focus and emphasis are notoriously difficult constructs to define, yet their role in affective expression is well documented in some speech communities (e.g., Irvine, 1990; Odlin, 1998, 1997), and they appear again and again in the grammatical descriptions of countless languages (Odlin, 2004b). A study by Schmid (1999) offers many examples where a structure-preserving translation from English to German is difficult if the translator's goal is to preserve the attitude of the speaker or writer. The difficulty is especially significant because the two languages share many of the formal devices conveying focus and emphasis, including inversion, left dislocation, and cleft sentences while yet they differ in their reliance on specific devices in specific contexts (cf. Doherty, 1999).

Bilinguals who are professional translators may err in their assessments of the pragmatics of affect in a source text, and the challenges for second language learners are certainly greater. They may judge as unacceptable translations that native speakers actually approve of, or they may overrely on some structures and show “syntactic conservatism,” not using the range of word order patterns licensed by the pragmatics of the target language (Aronsson, 2001; Håkansson, Pienemann, & Sayehli 2002, Plag & Zimmerman, 1998). As Callies (2002) suggests, even

advanced learners do not always have the competence to use target language structures in pragmatically appropriate ways even when the crosslinguistic similarities between the L1 and L2 are considerable.

The discussion in the preceding paragraph might seem to imply that there exists a strong constraint on the transfer of focus constructions. However, several studies show that the focus patterns of the native language often influence interlanguage patterns. In some cases, the transfer is partly negative as in an attempted cleft sentence noted by Odlin (2004b): He [Chaplin] say it were he som take the bred. This and other uses of som as a relative pronoun in a cleft structure occur in the writing of some Swedish learners of English but not in the English of L1 Finns (Odlin & Jarvis, 2004). Moreover, the same kind of cleft pattern with som occurred in the L1 writing of a Swedish control group. In some cases the transfer may be positive, as in the study of Aronsson (2001), who found frequent accurate uses of clefts by L1 Swedish speakers writing in English. Other studies of Scandinavian language contact likewise indicate transfer of word order and cleft patterns (Rosén, 2001; Westergaard, 2003). The discrepancies between the findings of these studies and those where transfer was not evident show the complexities of the issues and also the importance of methodology (cf. Callies, 2002).

The focus studies reviewed here look mainly at production and related metalinguistic behaviors such as translation and acceptability judgments. Speakers and writers of different languages may vary in what they judge important or how they speak or write their judgment, and such variations are relevant to the problem of conceptual transfer. No less crucial, however, is what listeners or readers understand of target language discourse, yet comprehension, the silent partner of production, has gotten less attention in studies of affect in conceptual transfer. However, a study by Graham, Hamblin, and Feldstein (2001) indicates that native and nonnative speakers differ in how they assess the emotions in the voices of English speakers, and the two groups of nonnative speakers (Spanish and Japanese) did not closely resemble each other. Such results thus indicate transfer involving comprehension. It is conceivable that similar kinds of receptive transfer will some day be documented in lexis and focus constructions.