Contact Linguistics. Chap

II. 5. 2. The concept of ‘transfer’ in group SLA.

We saw earlier that SLA researchers conceive of L1 influence primarily in terms of ‘transfer.’ We also noted that the term is used by some in a very broad sense, to refer to all forms of L1 influence (Odlin 1989:27). But most studies of transfer focus on only one type of L1 influence, that resulting from learners’ perception, whether conscious or not, of some similarity between L1 and L2 elements or structures.

As we saw earlier, Andersen (1983) refers to this kind of influence as “transfer to somewhere”, noting crucially that it can result in both successful and unsuccessful imitation of the TL (positive and negative transfer respectively). But L1 influence also occurs frequently where no similarity appears to exist between L1 and L2, on the basis of which learners can make “interlingual identifications.” Some have referred to these kinds of L1 influence as “transfer to nowhere” (Kellerman 1995) or “blind transfer” (Kean 1986). Kellerman cites examples from Slobin (1993) such as the preference for use of progressives rather than preterites in the L2 English narratives of Punjabi learners. He suggests that this is due to transfer of the Punjabi use of imperfectives in narratives, and to the fact that Punjabi is aspect-dominant. But it is arguable that this and the other examples Kellerman provides are actually cases of transfer to somewhere, based on misidentification. In this case, the English progressive is assigned functions associated with the Punjabi imperfective. This then is not a true case of transfer to nowhere. But the notion itself is worth pursuing further.
The main idea behind the notion of “transfer to nowhere” is that learners find certain aspects of the TL more difficult to acquire if they have no counterparts in the L1. Of course, not all such features pose difficulty. Some of them are transparent and hence easy to acquire, such as the singular/plural distinction in English, which are learnt by speakers of languages like Chinese, which do not mark the distinction morphologically. However, those differences that do constitute a barrier can have consequences for both learning and use of the TL. In attempting to use the TL, learners may resort to strategies based on their L1. It is such strategies that are apparently being associated with “transfer to nowhere.” A possible example of this might be the strategy Andersen (1990:62) refers to as “relexification”, though Andersen doesn’t make this connection himself. The strategy is subject to the following “relexification principle”:

When you cannot perceive the structural pattern used by the language you are trying to acquire, use your native language structure with lexical items from the second language.

In other words, this suggests that learners may use abstract L1 syntactic patterns as a template into which they insert L2 words. Andersen points out that this strategy may be more common in “natural” SLA settings, especially where access to L2 input is restricted. This appears to be quite true. For instance, Mesthrie & Dunne (1990:35-37) discuss cases of “transfer of relativization strategies from [Indic and Dravidian] substrates” in South African Indian English (SAIE). Indic languages employ a correlative construction with a prenominal relative clause, as illustrated by example (32) from Gujarati.

(33) Je vepari@ mari@ sathe avyo, te vepari@ Harilal ka bha@i@ che

CORRELAT. businessman me with came that businessman Harilal of brother is
“The businessman who came with me is Harilal’s brother”
[Lit: Which businessman came with me, that businessman is Harilal’s brother]

The same strategy is found in the L2 English employed by older speakers whose L1 is an Indic language. This is illustrated in (33)

(34) Which-one I put in the jar, that-one is good

“The ones [i.e. pickles] I put in the jar are the best”

Dravidian languages, on the other hand, employ a pre-nominal external relative clause without a relative pronoun, as illustrated in example (34) from Tamil.

(35) Van¡n¡en@e at¡icca tacca@n cenneki po@na@n

washerman.acc beat.past.rel. part carpenter.nom Madras.dat go.past.3sg.masc
“The carpenter who beat the washerman went to Madras”
[Lit. The washerman-beat(ing) carpenter went to Madras]

Again, a similar strategy is found in the L2 English of Dravidian speakers, as in the following example.

(36) People who got [working for them] sons, like, for them nice they can stay.

“It is nice for people who have sons [who are] working for [the company], since they are allowed to stay on [in the barracks].”

The SAIE relative structures, of course, are quite different from relatives in other varieties of English, though it might be argued that they are similar to structures that English would allow.

Interestingly, similar cases of syntactic retention can be found in the code-mixing behavior of bilinguals who are not very proficient in the L2, such as the Japanese-English bilinguals studied by Nishimura (1986) whom we discussed in Chapter 5. Compare also the retention of Japanese verb-final order and other features in the versions of Hawaiian Pidgin English used by Japanese immigrants in Hawaii, as in the following examples from Bickerton (1981). Japanese items are in italics.

(37) a. mista karsan-no tokoro tu eika sel shite

Mr. Carson-POSS place two acre sell do
“I sold two acres to Mr. Carson’s place.”

b. da pua pipl awl poteito iit.

“The poor people ate only potatoes.”

Strategies like these are not quite the same as the cases of word-order change we discussed earlier, in the IL of German learners of L2 English, such as the following example, repeated here for convenience.

(38) I would suggest him to go.

Cases like this do seem to involve transfer to somewhere, in that the English verb has been assigned the subcategorization properties of its German counterpart empfehlen ‘suggest.’ It is possible that this explanation also applies to the cases of so-called “transfer to nowhere” that we just discussed. In other words, these may involve the same kinds of “projection” of syntactic properties of L1 categories onto L2 elements as occurs in (38), and may simply be a more complex case of “transfer to somewhere.” (See Chapter 9, section 9.2 for discussion of similar kinds of “transfer” in creole formation.)

Another strategy that might be viewed as an instance of “transfer to nowhere” is the use of L1 content or function morphemes to convey meanings for which the L2 means of expression has not (yet) been acquired. We saw examples of this in the early interlanguage of German-speaking learners of L2 English, who use words like brills for ‘eyeglasses’ and German plural inflection as in doge ‘dogs.’ In all these cases, learners seem to follow the maxim; ‘When in doubt, fall back on L1 knowledge.”
It is clear then that L1 influence can manifest itself in various ways, and typically involves transfer to somewhere, though there may be a role for transfer to nowhere. This notion, however, remains tenuous and has not gained wide acceptance among SLA researchers.


II.5.3. Transfer as psycholinguistic process.

We noted earlier that the concept of transfer refers essentially to a psycholinguistic process by which learners manipulate L1 and L2 inputs to restructure their IL grammar. In cases involving transfer to somewhere, learners project the categories of their L1 onto L2-derived forms that have become part of their ILs. In other cases, learners try to compensate for unfamiliarity with certain TL structures by employing L1-based strategies that have no models in the TL.

This restructuring involves specific strategies and processes through which learners exploit L1 knowledge in creating the IL grammar. The strategies include:

(a) Retention of overt L1 elements in the IL, when L2 counterparts are unknown or non-existent.

(b) Retention of L1 structural patterns with insertion of L2 vocabulary (so-called “relexification”).
(c) Retention of L1 elements as substitutes for L2 counterparts perceived as similar (e.g. substitution of English /u/ for French /u/).
(d) Covert retention of L1 semantic categories via reanalysis/ reinterpretation of l2 forms that learners identify with L1 forms (e.g., reanalysis of after as a marker of perfect aspect in Irish English).

The first type of retention, (a), involves “transfer to nowhere” and by definition results in a departure from the TL (negative transfer). Type (b) can involve transfer either to somewhere or to nowhere, while types (c-d) involve transfer to somewhere. These three types (b – d) may have either positive or negative results in terms of accurate imitation of the TL.

Transfer, then, can be seen as the process by which certain L1 categories and structures are retained and projected onto L2-derived forms. It should also be borne in mind that L1 categories and structures may not be replicated exactly in the IL grammar. Learners may create compromises between L1 and L2 grammar, or other innovations that have no exact counterparts in either of the source languages.
Understanding these processes is a prerequisite to understanding the linguistic constraints on these types of L1 influence. The constraints have to do with the interlingual identifications learners make between L1 and L2 elements, based on the degree of congruence between these elements. Principles such as the Transfer to Somewhere Principle and the Relexification Principle can in fact be rephrased in terms of congruence-based constraints like the following:

29. 
    L2 structures that are highly congruent with those in the L1 will be acquired more easily (and successfully) than those that are not.
    
30. 
    L2 forms that are partially congruent with or partly similar in semantics or function to L1 forms will tend to be reanalyzed on the model of the latter.
    
31. 
    Certain L2 structures or elements that have no counterparts in the L1 may be difficult to learn. Learners may simply ignore such structures or employ L1 strategies by way of compensation.
    

Whether one interprets these as constraints on L1 influence or constraints on acquisition of L2 features is really a matter of perspective. Congruence-based constraints can also be interpreted in terms of markedness relations between L1 and L2 elements, as noted earlier.

We’ve suggested some ways in which L1 properties and elements may end up in the IL of individual learners. But under what conditions are such elements selected as part of a new contact vernacular? The answer to this lies primarily in social factors, which we can consider now.


II. 6. Non-structural factors in language shift.

Contact vernaculars like the new Englishes, Taiwanese Mandarin, etc., all seem to have arisen in settings characterized by limited interaction between native speakers of the TL and the groups learning it as an L2. In general, the TL is taught in schools, and earlier learners’ versions of it then become targets of acquisition for later learners. Since native speakers of the original TL are typically in a small minority and have higher social status, only the more educated and elite sections of the community have access to native TL models. Hence they tend to learn closer approximations to the TL, which may become the basis of a new local standard variety. The fact that most members of the community interact primarily among themselves rather than with native speakers of the TL means that the contact variety of the TL itself becomes the primary target of learning. Indeed, even teachers of the TL in the schools may speak a “local” version of it. These are perhaps the most important factors in the eventual preservation of features due to simplification, L1 influence etc. in the new vernacular.

Siegel (1997) proposed certain “reinforcement principles” that guide the selection process in these cases. In the first place, the greater the linguistic homogeneity of the population, the more likely it is that similar strategies of L1 retention etc will be found across the shifting group. Irish English, of course, meets this criterion well, since only one substrate language was involved. In the case of Singapore English, the numerical dominance of speakers of Chinese dialects ensured that the IL’s of these groups exerted most influence on the outcome. This said, it is of course impossible to predict exactly which features will be adopted, and which discarded. Siegel (ibid.) suggests that the key factor in this case is the salience of competing variants, as determined for instance by their frequency and/or transparency. However, we still know relatively little about the role of these or other factors. Some light may be shed on this by a comparison with processes of dialect leveling or koiné formation (Siegel 1985; Kerswill & Williams 2000). The factors that favor selection of certain variants over others in those situations are relevant here as well. (See discussion in Chapter 3, section 10.)
Certain differences in the patterns of contact between indigenous and L2 groups also seem to promote differences in the outcomes of shift. For example, in addition to schooling, migratory labor seems to have played an important role in the emergence of Irish English (Odlin 1997). Many speakers of Irish traveled either to parts of Britain or to English-speaking towns in Ireland to obtain work. This pattern of migration dated back to the 15^th century, and reached a peak in the late 18^th century when massive migrations to Britain occurred. On the one hand, most workers from Ulster tended to go to Scotland, thus reinforcing the establishment of Ulster Scots in that area. On the other hand, southern Irish tended to migrate to southern parts of England or to the English settlements in eastern Ireland, thus reinforcing the Southern English influence on southern Irish English. There was also a great deal of trade between southern Ireland and England in this period (Kallen 1997:146). The greater degree of contact with native English speakers in the Irish case seems to have resulted in Irish English being closer to its superstrate sources than Singapore English.
The persistence of bilingualism within the shifting group is another important factor in language shift. In both 18^th – 19^th century Ireland and 20^th century Singapore, the majority of those who acquired the TL maintained their ancestral language(s) as well. Odlin (1997a:4-5) suggests that there were large numbers of illiterate bilinguals in 19^th century Ireland, judging from the figures in the 1851 census. It is reasonable to assume also that childhood bilingualism was quite common, and that bilingual children played a role in the regularization of Irish English grammar. These factors would also have favored the retention of Irish features in the English of such speakers. Support for this comes from the fact, noted earlier, that Irish features are most pronounced in those varieties of contemporary Irish English spoken in areas that recently had or still have significant numbers of bilinguals. The Singapore situation is also characterized by high degrees of bilingualism, which may explain the pervasive substratal retentions in that variety.
No doubt there are many other factors relating to the community settings, demographics, patterns of interaction, etc, that played a role in these outcomes of shift. Gal’s (1979) pioneering investigation of the shift from Hungarian to German in Oberwart (Austria), explores some of these factors. She attributes the shift to several causes, including economic change, social mobility and opportunity, changing social network structures and speakers’ choices of social identity. All of these socio-cultural factors offer fruitful grounds for research on the causes and outcomes of shift.
Gal’s point about choices of social identity is particularly germane to the case of indigenized varieties. The preservation of distinctive characteristics in these languages has much to do with their value as symbols of national or ethnic identity for their speakers. They are often acquired as first languages, and serve as the primary means of everyday interaction, or in some cases as ‘link’ languages across ethnic groups, as in Singapore. Hence they have become associated with shared community membership and belonging. This has resulted in a range of problems concerning the status of these vernaculars, especially in cases where they co-exist with a lexically-related standard variety. Whether we focus on African-American English, Singapore English or Taiwanese Mandarin, the sociolinguistic reality remains the same. The vernacular varieties exist in a diglossic relationship with the standard dialects, leading to ambivalence both in attitudes and in public policy toward them. Studies of Irish English in Belfast (Milroy 1987; Milroy & Milroy 1992) reveal a tug-of-war between the status-oriented ideologies that confer prestige on the standard variety, and the identity-oriented values that confer legitimacy on the non-standard vernacular as a symbol of group cohesion. The pattern extends to all cases of unequal status between standard and non-standard varieties.
Our focus has shifted here from diachrony to synchrony, which links the study of shift situations to the broader sociolinguistic investigation of contact situations in general. The synchronic sociolinguistic study of indigenized varieties is still somewhat new, but it can contribute much to our understanding of the socio-cultural contexts of language contact. From a more practical perspective, such research can also address the problems associated with diglossia and its consequences for education, social opportunity and public planning.


II. 7. Questions of classification.

Where exactly do cases of shift fit into a typology of contact vernaculars? The answer to this is far from straightforward, since there is little consensus on what such a typology should look like in the first place. Even generally accepted categories such as ‘pidgin’, ‘creole’ and ‘bilingual mixed language’ are subject to great controversy. The category of ‘outcomes of shift’ is similarly vague and malleable. These outcomes range from fairly close approximations to a TL, to highly divergent versions of it. The degree of change in some of these cases has led researchers to treat them as close to creoles, which share similar processes of grammar restructuring.

For instance, Gupta 1992:342) refers to the formation of Sing. E. as a case of “creolization.” Similarly, Ho & Platt (1993:1) refer to Singapore English as “creoloid.” They argue that it “shares many of the features of a creole, e.g., serial verb constructions, variable marking of past tense, variable occurrence of articles and replacement of articles by other items, variable copula and auxiliary be.” Given all these characteristics, it is perhaps not surprising that Sing. E should be categorized as creole-like. But this approach assumes that there is a set of structural characteristics that is definitive of creole status, when in fact there isn’t. It is quite true that creoles share the characteristics that Ho & Platt list. But these are not unique to creoles. Rather, they result from processes of change and restructuring that creole formation shares with other cases of language contact, especially those involving natural SLA. If vernacular Sing. E. has serial verb constructions, for instance, it is because the L1’s of the learners of English in this contact situation (particularly Chinese dialects) have serial verb structures which were incorporated into Sing. E. via L1 retention, just as West African serial verb constructions were incorporated into Caribbean and other creoles. (See chapter 9.) On the other hand, the fact that the outcomes are not identical in all cases suggests that differences in the linguistic inputs and in the social settings play a significant role in determining the nature and extent of change that takes place.
One way to classify the outcomes of shift might be to place them on a continuum ranging from cases of relatively successful SLA at one extreme, to cases of radical creole formation at the other end. The chief criterion for placement would be the degree of L1 influence and perhaps simplification in the contact variety. Indigenized varieties would occupy the mid-range of this continuum, though they are by no means a homogeneous group.
Contact varieties like AAVE and Irish English might be placed closer to one end (say the left) on the scale, than varieties like Sing. E. Creoles also pose a problem for placement, since they too fail to constitute a homogeneous group. I will suggest in Chapter 9 that we make a broad distinction between “radical” creoles like Sranan Tongo and “intermediate” creoles such as Bajan (vernacular Barbadian English). The former would be placed on the extreme right of our continuum, while the latter would fall somewhere to the left of these, along with Singapore English.
Figure 1 is a rough sketch of the proposed continuum.


Figure 1: A continuum of outcomes of language shift.

Less L1 retention etc. <-------------------------> More L1 retention, simplification, etc.

Advanced Somewhat Highly Intermediate Radical

SLA indigenized indigenized creoles creoles

Native-like L2 AAVE, Irish E. Sing. E. Bajan Sranan

This is not meant to be an exhaustive or definitive classification of outcomes of shift. As already noted, there is no clear set of criteria that could be used to establish such a typology for contact outcomes in general. We will return to this issue in subsequent chapters, when we discuss pidgins and creoles.

Section III. First language attrition and death.


III. 1. Introduction.

Our discussion of language shift so far has focussed on the changes that occur in a TL during a process of group learning. But it is of equal interest and theoretical import to consider the changes that a speaker’s or group’s ancestral language (AL) can undergo as a result of shift to a new language. In many cases, the acquisition of a new language promotes the gradual abandonment of the AL, leading to increasing attrition and eventually the obsolescence and death of the latter. The processes of language decay and loss have traditionally been studied within the subdiscipline known as Language Death Studies, which has evolved in recent decades into a somewhat autonomous field of inquiry. However, as we will see, the phenomena involved in language attrition and death are similar to those found in many other cases of contact discussed throughout this book. Moreover, the social forces and changes that motivate language death are the same as those that promote contact-induced changes in language structure and behavior in general.

As Brenzinger & Dimmendaal (1992:3) points out, “a language is considered to be extinct when there is no longer a speech community using [it].” Gradual language shift is the primary, but not the only cause of language death. Sometimes languages become extinct because all their speakers die or are killed quickly, as in the case of Tasmania. In other cases, a group may suddenly abandon its language as a strategy of self-defense in the face of extreme repression, as happened with Lenca and Cacopera in El Salvador (Campbell & Muntzel 1989:182). But the vast majority of cases of language death are due to gradual abandonment of an AL.
Roughly half of the world’s known languages have disappeared within the last 500 years (Sasse 1992a:7). Hundreds more are in danger of dying. Most of these will probably not be remembered in the way in which long extinct languages like Gothic, Hittite, Sumerian etc. are. Great effort is being made to at least preserve records of these endangered languages, and in some cases even revitalize them. But the sad truth is that many if not most of them will be lost to us in the near future. The social and other external forces promoting such loss are often too powerful to counteract. Discussions of current attempts at language revitalization in various situations can be found in Fishman (1991; 2001).


III.2. External (social) factors in language death.

As Sasse (1992a:10) has pointed out, there are three general aspects of the study of language death. The first two focus on the macro- and micro-level social factors that lead to language obsolescence. The third focuses on the linguistic consequences for the dying language. Let us briefly examine the first two. Fortunately, there are several detailed studies of dying languages by scholars like Dorian (1981) on East Sutherland Gaelic, Sasse (1991) on Arvanítika in Greece, Gal (1979) on Hungarian in Austria, and others. These have given us a great deal of insight into the social ecology of language death.

III.2.1. Macro-level social factors.

Communities undergoing first language attrition tend to be minority or at least subordinate groups who have come under the control of larger or more powerful ones. Some of the factors leading to this kind of situation were discussed in Chapter 1 (sections 4.2 and 6.2) and Chapter 3 (section 8). They include the colonization of smaller ethnic groups by larger ones, and the formation of larger national polities that relegate some groups to subordinate status. Some of these factors were also discussed in Section II. 6 with respect to the shift from Irish to English in 19^th century Ireland.

Within such conglomerates, a variety of social forces come into play to compel the minority groups to assimilate to the dominant language and culture. The cultural, political and socioeconomic superiority of the dominant group is one such factor. This is usually reflected in the lower status ascribed to the minority language, often by language policies that stigmatize it and deny it a place in education, politics and other important spheres of life. In addition, increasing modernization, urbanization and economic changes have made it difficult for minority groups to retain their autonomy. Finally, the increasing availability of modern transportation and the reach of supra-regional communication media have made it difficult for previously isolated groups to avoid the pressure of the dominant language.
In the face of such pressure, some groups may deliberately decide to give up their AL in favor of a more prestigious or socially advantageous language. Brenzinger (1992b:213) mentions the case of the Yaaku of Kenya, who apparently held a public meeting in the early 1930’s and decided to abandon Yaaku and teach their children Maasai from then on. Most cases of shift, however, seem to involve an unconscious agreement that is slowly forced on the community by social forces within the community itself. Before we consider these internal social factors, let us consider the stages involved in the attrition and eventual death of an AL.


III. 2.2. Stages of Language Attrition and Death.

As Batibo (1992:90-92) informs us, language attrition occurs in several stages, which overlap to varying degrees. There is first a stage of monolingualism in the AL, followed by a period of growing bilingualism with the AL dominant. At this point, the AL is used in most kinds of in-group interaction, in domains like the family, neighborhood, etc. The L2 is used for wider inter-group communication. Stage 3 is a period of continuing bilingualism during which more and more speakers adopt the L2 as their primary language. Both stages 2 and 3 involve varying degrees of diglossia, allowing for code-switching behavior. The gradual breakdown of diglossia, wherein the L2 intrudes more and more on the domains of the AL, signals the beginning of complete shift. In the fourth stage, the members of the community display more limited knowledge and production of the AL. The final stage is the complete replacement of the AL by the L2. Traces of the AL often survive in the newly-acquired L2 as substratum features, as we saw earlier in the case of Irish English and other indigenized varieties. The five stages are not discrete, since different speakers shift toward the L2 to different extents at different times.

To summarize, the chief symptoms of first language attrition include a rapid decrease in the number of speakers, reduction in domains of use, and gradual loss of competence on the part of a growing number of speakers. The first two symptoms are more sociolinguistic in nature, and might be referred to as “quantitative attrition.” The third symptom has to do with the linguistic consequences of reduced competence, and might be referred to as “qualitative attrition.” This distinction is similar to that made by Rickford (1987:34) between “quantitative” and “qualitative” decreolization, which involves the gradual incursion of a standard language into the structure of a lexically-related creole language.


III. 2.3. Sociolinguistic factors within the shifting community.

Sociolinguistic approaches to language death are particularly concerned with the factors that come into play in quantitative L1 attrition. These factors are not different from those that regulate language choice in bilingual communities in general, which we discussed in Chapter 4. It is their specific configuration that promotes shift in some cases and inhibits it in others. The overall sociolinguistic structure of the community, its patterns of language usage, its size and its ideology toward language all determine the extent to which its ancestral language is prone to decay. One overriding factor that promotes language maintenance is a high degree of vitality in an ethnolinguistic group. Giles et al (1977:308) defined group vitality as “that which makes a group likely to behave as a distinctive and collective entity within the inter-group setting.” The higher the degree of vitality, the greater the degree of loyalty to the AL is likely to be. (See also the discussion in Chapter 4, section 3.4.) On the other hand, the pursuit of socio-economic opportunity and social advancement, in concert with socio-psychological pressure from the dominant group may lead to negative evaluation of the AL, thus providing strong motivation for shift.