15.1.1 What is attrition?

Broadly defined, and as used here, language attrition refers to the temporary or permanent decline of language skills, knowledge, and/or use in individuals. It is both a process and a potential outcome of the dynamic sociolinguistic conditions of language contact in which individuals may find themselves. De Bot and Weltens (Reference de Bot and Weltens1985) and Van Els (Reference Els, Weltens, de Bot and Els1986) distinguish four bilingual situations in which language attrition occurs, each one marking a different combination of the language that is lost (L1 or L2) and the environment in which language loss takes place (L1 or L2 environment). This chapter is mainly concerned with L1 attrition that occurs in L2 environments, such as the deterioration of the native language of immigrants. L1 attrition is a non-pathological, commonly occurring phenomenon in individuals, but in situations of unstable contact it may create conditions for language shift at the community level (Myers-Scotton Reference Myers-Scotton2002b). From a linguistic point of view, erosion in the immigrant L1 is a result of intensive language contact; that is, changes that occur in the L1 “would have been less likely to happen if it were not for the contact situation” (Thomason Reference Thomason2001:62). Nevertheless, contact-induced erosion of L1 knowledge is rarely a “total phenomenon” (Clyne Reference Clyne, Fishman, Tabouret-Keller, Clyne, Krishnamurti and Abdulaziz1986); rather, it is a gradual and selective process in which distinct aspects of the waning language can be affected differently (Seliger Reference Seliger, Seliger and Vago1991). Some of the changes appear as loss, reduction, or replacement of certain L1 features, while others seem to enrich the L1 system through innovation, addition, and borrowing from the L2 (Gal Reference Gal and Dorian1989; Sharwood Smith Reference Smith, Michael, Hyltenstam and Obler1989). Indeed, L2 influence on L1 as a major mechanism in the attrition process has been the focus of much recent work on adult bilinguals (Pavlenko Reference Pavlenko2000; Gürel Reference Gürel2004).

More narrowly defined, attrition is the forgetting or “loss of aspects of a previously fully acquired primary language resulting from the acquisition of another language” (Seliger Reference Seliger, Ritchie and Bhatia1996:606, emphasis added). Assuming that attrition proper presupposes complete L1 knowledge before the onset of bilingualism and subsequent L1 loss, the use of the term seems more felicitous in reference to adult bilinguals than young children. There is ample evidence that most healthy, monolingual children attain mature and stable competence in the basic structures of their native language by the age of five; mastering the most complex constructions may take a few additional years, while vocabulary acquisition continues for another decade or so (Aitchison Reference Aitchison1996). Strictly speaking, then, attrition may not take place before the age of linguistic maturation – after all, what has never been acquired cannot be forgotten. Such a narrow definition of L1 attrition may be difficult to apply to simultaneous and early bilingual children growing up with a minority L1 and majority L2, since the L1 may begin to erode before it has been fully mastered due to persistent dominance of the societal L2.

15.1.2 Incomplete L1 acquisition in children

In an attempt to mitigate the problem of terminological ambiguity, recent studies refer to bilingual L1 erosion in children as developmental attrition and/or as imperfect or incomplete acquisition (Kaufman and Aronoff Reference Kaufman, Aronoff, Seliger and Vago1991; Polinsky Reference Polinsky1995; Halmari Reference Halmari2005; Bolonyai Reference Bolonyai2007). Clearly, the distinction between forgetting and incomplete acquisition is an essential one; however, it is also important to recognize that the two processes can co-occur and reinforce each other, resulting in linguistic outcomes that may not be clearly assigned to one or the other process.

Incomplete L1 acquisition is fairly common in first- and second-generation minority children who acquire their L1 mainly as the home language while being immersed in a socially and functionally L2-dominant environment. The language dynamism associated with these children is regarded as a “special case of language acquisition and use” (Andersen Reference Andersen, Lambert and Freed1982: 86). What makes this acquisition–attrition process unique is that it involves “divergent change” (Sharwood Smith Reference Smith, Michael, Hyltenstam and Obler1989) in performance and/or competence and concomitant linguistic behavior that appears to violate “normal” – i.e. native, monolingual – linguistic and sociolinguistic norms. In most cases, changes and deviations in an L1 system that has not fully developed or stabilized can be attributed to the child’s restricted contact with the native language (i.e. when the quantity and/or quality of L1 exposure and functional use is limited) alongside intensive contact with L2 (i.e. when the predominant source of linguistic input and active use is L2). The onset of the process of contact-induced divergence is important because “the earlier a child’s L1 input becomes restricted and the more restricted the continued input is, the more likely the child is to show signs of incomplete acquisition in the areas of morphology, syntax, semantics, and lexicon, as well as phonology” (Halmari Reference Halmari2005:340).

This claim is in line with the view that children must be exposed to a “critical mass” of input, both in quantity and quality, during a certain period for language maintenance to be successful (Neisser Reference Neisser1984; de Bot and Clyne Reference de Bot and Clyne1989). If exposure to the minority language continues to be curtailed through puberty and adulthood, the individual may ultimately remain an incomplete learner (Polinsky Reference Polinsky1995) or imperfect speaker (Dorian Reference Dorian1981) of his/her native tongue. Indeed, several studies of second-generation adult speakers of heritage languages report a positive correlation between reduced access to the full, native L1 variety over time and the emergence of a qualitatively divergent, imperfect adult language system (Silva-Corvalán Reference Silva-Corvalán, Seliger and Vago1991; Polinsky Reference Polinsky1995; Montrul Reference Montrul2002). Undoubtedly, when an imperfectly learned variety of the heritage language is transmitted to subsequent generations, it is even more likely to undergo additional structural changes.

15.2.1 Language status

On the sociolinguistic level, asymmetry is most commonly implicated in the political and symbolic economy of the languages in contact. Linguistic inequality between a majority and a minority language is constructed on the dominant “linguistic market” that endows languages with different values and helps to (re-)produce asymmetrical power relations (Bourdieu Reference Bourdieu1991). Thus, in the context of minority language attrition among immigrants, the societal language is presumably more highly valued than the minority L1. Supported by legitimating language ideologies, the socially dominant language is typically recognized as the language of power, status, authority, prestige, and the beacon of socio-economic mobility. As linguistic capital, the immigrant language is rarely positioned to carry equal clout as the societal language. Of course, that is not to say that the immigrant language is always lacking currency or that possessing any two (or more) languages in one’s repertoire is not a valuable asset. In fact, the more linguistic capital that speakers accumulate, the more they should be able to profit from the hierarchy between languages (Bourdieu Reference Bourdieu1991). Yet, it is not an exceptional case when the symbolic value of ethnicity, cultural heritage, in-group solidarity, and emotional intimacy associated with a minority L1 fails to provide the incentive necessary for successful trans-generational language maintenance.

15.2.2 Language attitudes and use

It is also not uncommon for an asymmetrical sociolinguistic order to promote psycholinguistic imbalance between the L1 and L2, for example, in terms of speakers’ attitudes towards their languages. Parents’ language attitudes are generally considered crucial in children’s L1 retention. When parents view the native language as an integral part of ethnicity, regard it as a valuable economic asset, and/or a highly prestigious source of cultural capital, and its trans-generational retention important, they are more likely to use the ethnic L1 at home as the primary means of communication. Yet, numerous studies indicate that parental encouragement and prevalent L1 use notwithstanding, immigrant children who leave their native country at a very young age or are born in the host society may develop more positive attitudes toward the majority L2 than the minority L1 over a period of time (Hakuta and D’Andrea Reference Hakuta and D’Andrea1992; Kouritzin Reference Kouritzin1999; Young and Tran Reference Young and Myluong1999; Luo and Wiseman Reference Luo and Wiseman2000). For example, Hakuta and D’Andrea (Reference Hakuta and D’Andrea1992) investigated the relationship between language attitudes and use in Mexican-American adolescents of first- and second-generation immigrants in California. They found that teenagers who held positive attitudes toward L2 (English) reported using English more often. Self-reports of estimated proficiency in both L1 and L2 were predicted by their attitudes toward each language. Another study by Kaufman (Reference Kaufman, Ammerlaan, Hulsen, Strating and Yağmur2001) explored the role of attitudes in L1 attrition among children of Israeli immigrants residing in the United States. This study too argues for a powerful link between language attitudes and outcomes of bilingual contact by reporting that children’s low affective attachment to the L1 (Hebrew) and high integrative orientation toward the L2 (English) make the L1 particularly vulnerable to attrition.

15.2.3 Language and identity

While negative attitudes toward the immigrant language can play a significant role in its deterioration, L1 attrition, incomplete acquisition, or shift to L2 may occur even when speakers maintain strong and positive identification with the heritage language and culture (Bankston and Henry Reference Bankston and Henry1998). Due to sociolinguistic conditions brought about by migration and language contact, immigrant minorities’ conception of what it means to be, for example, Puerto Rican, or who counts as a legitimate speaker of Spanish, might change. Assumptions, taken for granted before migration, about a “natural” and intimate connection between ethnic language, identity, and culture may be revised such that the notion of retaining one’s ethnic identity while not speaking the ethnic language is perceived as normal. As a case in point, Zentella’s (Reference Zentella1997) study of Puerto Rican children growing up in New York demonstrates that a Puerto Rican identity for the youth born and/or raised in El Barrio does not depend on speaking Spanish. The study reports that given the “concrete reality” of immigrant life, this community “drew the boundaries of Puerto Rican identity wide enough to encompass the monolingual children of Puerto Rican descent” (Zentella Reference Zentella1997:54). Parents aim for their children to achieve a minimal level of Spanish comprehension; in fact, community norms define those with passive comprehension in one language and fluent speaking and understanding in the other language as bilinguals. Predictably, language choice patterns are often non-reciprocal in that parents may speak Spanish and children respond in English. Similar asymmetrical language choice patterns in inter-generational communication are well documented in the literature (Gal Reference Gal1979; Bentahila and Davies Reference Bentahila, Davies and Harris1992; Extra and Verhoeven Reference Extra and Verhoeven1998; Li Wei Reference Li Wei1994; Hlavac Reference Hlavac2000; Clyne Reference Clyne2003). Zentella further notes that children who lack minimal Spanish skills are accommodated through code-switches to English or translation both in peer and adult-child interactions. The function of switching between languages, however, goes beyond accommodation. It is “the most obvious expression” (Zentella Reference Zentella1997:271) of the multiple and complementary identities immigrant bilinguals forge for themselves as they (re-)define what it means to belong to an ethnolinguistic minority community within the political economy of a broader social context. Questions of how CS may be implicated in the linguistic processes and outcomes of children acquiring a minority L1 in a majority L2 setting is discussed next.

15.3.1 Sociolinguistic evidence

CS in language erosion can be distinguished from normal CS on a sociolinguistic basis. Unlike fluent bilinguals, whose linguistic competence enables them to use their languages in accordance with relevant sociolinguistic norms and situational characteristics, those undergoing L1 attrition may be forced to code-switch when it is not appropriate. Seliger (Reference Seliger, Ritchie and Bhatia1996:613) claims that, “[c]ode-mixing can be considered a precursor condition for primary language attrition when mixing begins to occur in contexts that are not motivated by external factors such as interlocutor, topic, or cultural environment.” Empirical evidence for this claim comes from a study of a Russian–English-speaking child’s speech prior to and after attrition (Turian and Altenberg Reference Turian, Altenberg, Seliger and Vago1991). Prior to attrition (3;0 to 3;7 years;months), the child appropriately used CS to address each family member in his or her respective L1, or for emphasis. By contrast, post-attrition data (at 4;3 and 4;4 years of age) show continuous CS to English, unrelated to interlocutor, topic, or context of conversation. In example (1), J’s mixing English words into Russian when speaking to a monolingual Russian speaker, A, is prompted by his lack of proficiency.

15.3.2 Psycholinguistic evidence

Psycholinguistic factors triggering CS include on-line lexical retrieval difficulties, fluency problems, and gaps in “an incomplete knowledge base” (Poulisse Reference Poulisse1999). Lack of availability or accessibility of certain words and structures can be reflected in the use of CS and borrowings “flagged” with pauses, hesitations, false starts, repetitions, fillers, inaccuracy, avoidance, and reformulations (Olshtain and Barzilay Reference Olshtain, Margaret, Seliger and Vago1991). While any bilingual may produce flagged CS, it is much less likely to be used by fluent bilinguals than those who are in the process of language attrition. Bilingual speakers, who use their languages with sufficient frequency, are generally able to control, access, and activate each language according to their communicative goals with great facility, accuracy, and fluency (Green Reference Green1998; Paradis Reference Paradis, Brend, Melby and Lommel2001). Nevertheless, occasional CS due to lexical retrieval problems is considered normal, since even fluent bilinguals rarely develop equal, perfectly balanced proficiency or pattern of use in their two languages (Grosjean Reference Grosjean1998). CS may also be prompted by a momentary, tip-of-the-tongue memory lapse, which, of course, occurs in bilinguals and monolinguals alike.

In L1 attrition and incomplete acquisition, reduced use and accessibility of L1 may result in frequent or involuntary CS to the stronger L2 (Turian and Altenberg Reference Turian, Altenberg, Seliger and Vago1991). Speakers use CS as a “crutch” in coping with permanent or temporary inaccessibility of specific words and complex or unstable structures (Zentella Reference Zentella1997). When CS is motivated by L1 erosion, it often co-occurs with pauses and other disfluency phenomena, appeals for help, and metalinguistic comments (e.g. “I forgot,” “I don’t know how to say it”) that explicitly signal a breakdown in communication. In (2), two English-dominant Hungarian-American children, ages eight and nine, talk about their favorite movies in their weaker L1. The child’s (CH1) difficulty finding the appropriate Hungarian word manifests itself in the use of repeated pauses, a filler, and finally CS to English, her stronger language.

Although it is often assumed that the main reason bilingual children use CS is to compensate for their linguistic deficiency, empirical research does not uniformly support this assumption. Zentella (Reference Zentella1997) examined the CS strategies used by five Puerto Rican children (aged six to eleven) in New York and found that CS for “crutching” (i.e. when the child did not know or remember a word) accounted for only about 14% of all instances of CS in the corpus. The majority of CS served conversational functions such as change in footing and clarification and/or emphasis, as shown in examples (3) and (4). Crutching occurred most frequently in CS to English, suggesting children’s dominance in English. Overall, however, the findings show that “even non-fluent children do less ‘crutching’ than most people assume” (99).

15.3.3 Linguistic evidence

From a grammatical perspective, CS has been characterized in the literature as a sophisticated, structurally coherent, rule-governed behavior that requires a great deal of bilingual competence. While there is little consensus with respect to the nature of linguistic rules or constraints involved in CS, most researchers agree that the degree of bilingual competence constitutes a source of variability in the type of CS that is produced (Weinreich Reference Weinreich1953; Clyne Reference Clyne1967; Myers-Scotton Reference Myers-Scotton1993a; Reference Myers-Scotton1997; Reference Myers-Scotton2002a; Muysken Reference Muysken2000). For example, studies based on both adult and child populations have found that unbalanced bilinguals tend to favor single-word and tag-like switches intra-sententially. The direction of these switches is typically from speakers’ stronger language to the weaker language. This pattern has been used to support claims that the less morpho-syntactic anchoring an alien element requires in the base language, the easier it is to use in CS.

Another CS pattern that is likely to occur in unbalanced bilinguals is alternational CS between sentences or speech turns (Poplack Reference Poplack1980). When speakers are constrained in their grammatical abilities in one of the languages, this CS style may help to avoid the production of longer and morpho-syntactically more complex stretches of speech. By the same token, intra-sentential CS involving larger segments and constituents (phrases, clauses) requires a high level of proficiency in both grammars, and therefore is, arguably, more likely to occur in the most fluent and balanced bilingual speakers. The following examples from Hungarian–English data may illustrate the distinction between intra-sentential word-level switching (5), phrasal-level switching (6), and inter-sentential turn-level switching (7).

It is important to realize, however, that the aforementioned correlations are probabilistic rather than absolute. For example, Bentahila and Davies’s (Reference Bentahila, Davies and Harris1992) study of two generations of Arabic–French bilinguals in Morocco reported a different patterning of CS. The older generation of bilinguals, who had an equally high command of both languages, used the socio-linguistically more acceptable inter-sentential CS more frequently, whereas the younger bilinguals, who were Arabic-dominant, used intra-sentential CS with more frequency. Additionally, there is some evidence that intra-sentential CS may be a more frequent pattern in CS involving an agglutinative language than CS involving a fusional language (Myers-Scotton Reference Myers-Scotton1993a; Halmari Reference Halmari1997; Clyne Reference Clyne2003). Indeed, it is incumbent on future research to systematically account for the exact ways in which sociolinguistic and typological factors may interact with psycholinguistic and cognitive (competence-related) factors to produce the actual type of CS.

Unlike “normal” bilingual CS, where the grammatical autonomy of each language is maintained and shows largely predictable structures, CS in L1 erosion is lacking in structural coherence, predictability, and well-formedness, as measured by community norms (Myers-Scotton Reference Myers-Scotton1997). Motivated primarily by the speaker’s linguistic needs rather than socio-pragmatic goals and linguistic constraints, CS showing attrition has been characterized as linguistic “intermixing” where the autonomy and integrity of the L1 system is no longer observed (Sridhar and Sridhar Reference Sridhar and Sridhar1980; Seliger and Vago Reference Seliger and Vago1991). According to Seliger (Reference Seliger, Ritchie and Bhatia1996:611), language mixing in attrition “leads to the nonobservance of language-specific constraints” on L1, and, eventually, the fusion or merging of the two grammatical systems in contact into a single system. Due to reduced accessibility of the L1 grammar and intensive contact with L2, the L2 linguistic system becomes a source of “indirect positive evidence” (Seliger Reference Seliger, Seliger and Vago1991), a model for imitation, or “copying” (Johanson Reference Johanson, Extra and Verhoeven1993). While this process is selective, perceived cross-linguistic congruence – on the basis of the speaker’s subjective assessment of equivalence between any given structure in L1 and L2 – has been claimed to have an important effect on determining how and what linguistic features of the L1 will undergo change (Johanson Reference Johanson, Extra and Verhoeven1993; Bolonyai Reference Bolonyai2000; Muysken Reference Muysken2000; Backus Reference Backus2005). As numerous studies have shown, L2 influence on the unstable L1 may manifest itself in various forms of language mixing, such as transfer, interference, transference, structural borrowing, selective copying, calquing, convergence, restructuring, creative innovation, incorporation of L1 into L2, covert CS, composite CS, or “third-system” innovation, among others. The following examples from the speech of young immigrant children in the United States illustrate the type of language mixing that could be considered as a sign of L1 attrition and/or imperfect acquisition (Bolonyai Reference Bolonyai1998, Reference Bolonyai2002, Reference Bolonyai2005, Reference Bolonyai2007).

Example (8) comes from a Hungarian–English-speaking, four-year-old child residing in the US. The utterance illustrates what has been referred to as covert CS, or convergence: “bilingual speech appearing in the disguise of monolingual speech” (Bolonyai Reference Bolonyai1998:23). In covert CS, all lexical material comes from one language, but abstract structural features from both languages converge in a composite grammatical structure, or “composite Matrix Language” (Myers-Scotton Reference Myers-Scotton, Grenoble and Whaley1998, Reference Myers-Scotton2002a; Bolonyai Reference Bolonyai1998; Schmitt Reference Schmitt2000). Composite grammatical structures are likely to be found in situations of L1 attrition/imperfect acquisition, when unbalanced, non-fluent bilinguals attempt to communicate in their linguistically unstable, weaker language. In (8), all the words are in Hungarian, although some of the structural properties of the utterance are copied from English. The word order is directed in part by Hungarian, in part by English. In line with Standard Hungarian rules, subject pro-drop is observed. Following English, however, the verb voltam “I was” precedes the complement meleg “hot,” which is in violation of the Hungarian rule for marking pragmatic emphasis. Also, the argument structure (Dative Experiencer Subject) of the Hungarian construction “to have hot” is replaced by the argument structure (Nominative Experiencer Subject) of its English equivalent “to be hot.” Thus, the monolingual surface form of the utterance is supported by a composite structure, with parts of abstract lexical structure from two linguistic systems combined in a way that, arguably, reflects the child’s subjective perspective on “interlingual equivalence” (Johanson Reference Johanson, Extra and Verhoeven1993) and structural compatibility between Hungarian and English.

Example (9) is taken from a study that examines the vulnerability of the Hungarian case system in the speech of six Hungarian-American children, ages seven to nine (Bolonyai Reference Bolonyai2002).

The example illustrates composite CS, where code-switched L2 lexemes are incorporated into the L1 system along with some L2-specific abstract structural properties. Specifically, English influence on Hungarian is noticeable in the use of semantic case morphology, which, in turn, indicates signs of restructuring at the conceptual–semantic level – in the mental representation of topological space, such as “container” (“in”), “surface” (“on”), and “proximity” (“to”). Unlike in English, in Standard Hungarian a classroom as an “imagined” place, associated with particular disciplinary knowledge, discourse, and identities (as opposed to a concrete place of a classroom), is conceptualized as a surface and therefore takes on case endings (e.g. sublative –ra “onto”). In (7), however, the English code-switched nouns for classes (P.E., music, art) receive an in case (illative –ba “into”), which marks them as a container – just as they are in English, where the distinction between the two senses of “class” is not grammaticalized but shows a partial overlap with Hungarian. According to the findings of the study, morphological case replacements were most frequent in the L1-specific structuring of topological space due to cross-linguistic influence from the children’s dominant L2/English on spatial “thinking for speaking” (Slobin Reference Slobin, Gumperz and Levinson1996).

Examples such as these are commensurate with the suggestion that “interface” structures, such as topological spatial expression – which require the integration of conceptual–semantic and morpho-syntactic knowledge – might constitute a vulnerable area in bilingual L1 attrition and imperfect acquisition. The hypothesis that cross-linguistic influence is most likely to occur in grammatical phenomena that involve the semantics–syntax or pragmatics–syntax interface has been in the forefront of more recent work within generative approaches to bilingualism and language contact (Hulk and Müller Reference Hulk and Müller2000; Sorace Reference Sorace, Howell, Fish and Keith-Lucas2000; Montrul Reference Montrul2002, Reference Montrul, Schmid, Kopke, Keijzer and Weilemar2004; Bullock and Toribio Reference Bullock and Toribio2004; Gürel Reference Gürel2004; Sorace Reference Sorace2004; Toribio Reference Toribio2004; Tsimpli et al. Reference Tsimpli, Sorace, Heycock and Filiaci2004; Bolonyai Reference Bolonyai2007). Studies of different types of bilingual development, such as bilingual L1 acquisition, adult L2 acquisition, and L1 attrition, have indicated that grammatical interfaces are particularly unstable and open to cross-linguistic influence, leading to optionality and variability in the production/interpretation of affected features (Sorace Reference Sorace2004). According to Sorace, interface instability in attrition/imperfect acquisition can be attributed to two main factors. First, interfaces are more complex than narrow syntax and therefore may be acquired late, partially, or never – that is, certain grammatical properties remain underspecified or “permanently indeterminate” (Reference Sorace2004:143) at the level of competence. Second, speakers may lack processing resources that are required to integrate both (morpho-)syntactic and conceptual (pragmatic, semantic) constraints governing a particular linguistic structure. Although cross-linguistic influence and CS often co-occur in bilingual L1 attrition/imperfect acquisition, the question of whether CS may be connected to the vulnerability of interface structures has received very little attention (see Bolonyai Reference Bolonyai2007).

Finally, an important, although sometimes neglected, aspect of L1 attrition/incomplete acquisition in the context of bilingualism is its potential for linguistic creativity and innovation (Gal Reference Gal and Dorian1989; Seliger Reference Seliger, Hyltenstam and Obler1989; Thomason Reference Thomason2001; Skaaden Reference Skaaden2005). Driven by linguistic need, sociolinguistic incentives, or perhaps ludic motivations for expressiveness, bilingual speakers (their limited L1 resources notwithstanding) may produce unconventional innovations that evidence lexical and structural creativity. While some of the innovations are intra-lingual, i.e. based on monolingual L1 resources, many others exploit L2 resources and/or involve CS. Examples (10) and (11) show innovative bilingual verb formation patterns found in the spontaneous speech of pre-pubescent Hungarian–English bilingual children.

Previous research has demonstrated that fluent Hungarian speakers use a verbalizer suffix (-ol) to integrate borrowed and code-switched words into Hungarian. According to Moravcsik (Reference Moravcsik1975), foreign verbs are borrowed into Hungarian as nominal forms; therefore, they must be verbalized. In the children’s data, however, the verbalizer may attach to unambiguous verb stems from English such as donate in (10). Since this construction (verb + denominal verbalizer + inflection) does not exist in either Hungarian or English but does occur in bilingual contact, it can be seen as an innovative structure. It suggests change in the semantic features of the verbalizer, which enables creative productivity in bilingual verb formation. In other cases, code-switched English verbs are inserted into Hungarian without a verbalizer, as in (11). This strategy may indicate that the English verb (“hear”) is treated as fully equivalent with a Hungarian verb – rather than a “foreign non-verb” needing to be nativized. Linguistic creativity in intensive language contact thus can be linked to the speaker’s assessment of perceived equivalence between L1 and L2, whereby linguistic asymmetries and boundaries that mandate the presence of the verbalizer in “normal” CS are being reconfigured. Of course, ultimately, as Gal points out, both innovation and loss are the consequences of an ongoing conflict and competition between cognitive, interactional, and social/symbolic forces: “in the midst of diminishing use and input from Hungarian, young speakers must nevertheless use that language to communicate effectively” in their family or community networks (Reference Gal and Dorian1989:330). That is, linguistic creativity can be seen as a response to the interactional demand of family and ethnic community networks.