"I'm forgetting both": L1 maintenance and code switching in Finnish-English language contact.

Abstract

This article is a case study investigating the maintenance of Finnish in the language of two young Finnish-American bilinguals. The main source of data is a 90 mins audio-taped conversation between the research participants, which was recorded in Finland after they had lived eight years in the United States and become fully fluent in English. Three main questions are addressed: (1) Are there signs of first language attrition or change from Finland-Finnish patterns in the area of morphosyntax? (2) Are there signs of imperfect acquisition of Finnish lexicon? (3) Can the research participants' codeswithing patterns be seen as indications of attrition in Finnish morphosyntax and/or imperfect acquisition of Finnish?

Key words

codeswitching

Finnish-English

L1 maintenance

**********

The data indicate that the complex Finnish morphosyntax is used with few deviations from the Finland-Finnish patterns; however, there are clear signs that, because of the lengthy stay in an English-speaking environment, lexical deviations from Finland Finnish do occur. While both research participants resort to codeswitching, this is motivated not by the attrition of Finnish morphosyntax but by lexical gaps in registers requiring special vocubulary, as well as by multifarious discourse-related goals, earlier well documented in the codeswitching literature. The article underscores the heightened bilingual consciousness in the language of these two Finnish-English bilinguals. This bilingual consciousness often leads to metalinguistics discussions and conscious repair phenomena when either an English word is inserted into otherwise Finnish discourse, or when the speaker accidentally produces an ungrammatical Finnish form.

While the language of the research participants shows some change from Finland Finnish, this change is mainly manifested in lexical innovations and not in the morphosyntactic patterns. While codeswitching can be used as an avoidance strategy when a Finnish word has not been acquired or is not accessed as quickly as its English counterpart, codeswitching in these data is mainly used as an interactional strategy.

1 Can codeswitching patterns tell us about language attrition?

A large body of literature has accumulated on the topic of primary language attrition (see, e.g., Cohen & Weltens, 1989; Isurin, 2000: Lambert & Freed, 1982; Seliger, 1996; Seliger & Vago, 1991: Skaaden, 1999; Weltens, de Bot, & van Els, 1986). This attrition, if not pathological attrition as in cases of aphasia and dementia in monolingual speakers, is triggered in situations of language contact--the speakers' L1 starts to fade away when it comes into prolonged and obtrusive contact with their L2 (e.g., de Bot & Clyne, 1994; Dorian, 1981, 1982: Grenoble & Whaley, 1998; Polinsky, 1994, 1995, 1997; Silva-Corvalan, 1991; Williams, 1999: for more general accounts on language change induced by language contact, see, e.g., Croft, 2000; Silva-Corvalan, 1994; Thomason & Kaufman, 1988). In a language contact situation, the mixing of the L1 and L2 by the bilingual speaker may be perceived as a sign of incipient L1 loss (Halmari, 1992: Seliger, 1996, p. 610); however, in this case study I will show that while language loss--in a situation where the bilingual speakers, over an extensive period of time, are exposed to their L1 mainly as the home language--is certainly always a lurking possibility, and while its onset may be detected in codeswitching patterns, codeswitching per se is not a sign of language attrition. I argue that much of the codeswitching into English, present in my data, does not reflect L1 attrition and is used as an interactional strategy.

1.1 Attrition or "normal" codeswitching?

Seliger defines the difference between language switching as an indication of language loss and language switching motivated by other factors than L1 erosion in the following way:

  ... the language behavior of an attriter might resemble normal  codemixing and to the observer might not indicate that there is  anything amiss linguistically, because all of the speakers in    the  situation are switching back and forth between the two    languages. The  difference is that the codemixing bilingual can    switch to either  language with approximately equal facility    without any disruption in  fluency or grammatical accuracy.    (Seliger, 1996, p.610) 

If this quotation is taken at its face value, language dominance could play an important role in the directionality of codeswitching. To simplify a complex issue, it may be argued that, theoretically, codeswitchers switch equally often into both directions (depending, of course, on speech-situational factors); attriters (Seliger's term) prefer switching to their L2; and, L1-dominant bilinguals prefer switching to their L1. Naturally, a codeswitcher may become an attriter over a period of time, and an attriter may gain back the competence in the L1 that has started to erode. Bilingualism is never guaranteed to be a stable state of affairs, and child bilingualism is especially vulnerable to changes in the competencies of the two languages. Therefore, if L1 maintenance and a more or less balanced competence in the child's two languages are the goals, it is important to monitor the codeswitching patterns with the following question in mind: Are these patterns reflective of incipient attrition and/or only partial acquisition of the L1? This fundamental question breaks down into two more specific questions: (1) Have changes occurred in the codeswitching patterns? and, (2) Do the switched segments internally adhere to the grammatical constraints of their language? What may complicate the assessment of the grammaticality of the L1 with regard to the latter question is the fact that especially in large immigrant communities the L1 may change, developing to exhibit "new norms" that look like violations of monolingual grammar. Since this article is concerned with bilingual speakers who belong to a very small immigrant community consisting mainly of members of an isolated immigrant family, and since within this family the use of the norms of monolingual Finnish (i.e., Finland Finnish) is an explicit (albeit it often subconscious) goal, I will, instead of an already changed immigrant L1 variety, use the monolingual L1 variety as my point of comparison.

An important point to elaborate on with regard to the quote above is its claim that "the codemixing bilingual can switch to either language with approximately equal facility without any disruption in fluency or grammatical accuracy" (Seliger, 1996, p. 610). In actuality, this is rarely attested in bilingual communities; switching equally easily and equally often in the direction of both languages is not common in naturally occurring bilingual data, no matter how well represented the two languages are. (1) Processing and register-related issues have a strong effect on bilingual performance. In the same way as monolingual speakers are rarely as fluent in all the linguistic registers of their one language, bilinguals do not have equal command of different speech situations in both of their languages. One language may be more dominant in one situation; the other language in others. Thus, language switching in itself does not mean that there is attrition in either language (Pfaff 1979; Seliger, 1996). We all know how a language we have not used for a long time--but have once known well--will take a few days to activate to a level which allows fluent performance. If a language is not activated, access issues may lead to codeswitching.

Language purists often frown upon codeswitching, easily attributing its occurrence to attrition. I am not denying that this may often be the case. However, as Fuller (1999, p. 550 and references therein) points out, the motivations for codeswitching can also be traced back not to "the lack of knowledge of the structure of the ML [Matrix Language]" but to "social and psychological reasons of identity, denotation, and connotation" or to gaps "in access to semantically or pragmatically appropriate words in the ML."

In child bilingualism, with which this article is mainly concerned, another important reason for frequent codeswitching may be the fact that the child has been cut off from rich input in her or his L1 at a point when the L1 had not yet developed to what would be an adult target norm. I will address this issue--the issue of imperfect acquisition--in the next section.

1.2 Attrition or imperfect acquisition?

Critical for the issues in this paper is the distinction between L1 attrition and imperfect acquisition of the L1. (2) By definition, the issue of imperfect acquisition of the L1 pertains mostly to childhood, not adult, bilingualism.

However, before discussing imperfect acquisition in child bilingualism, I want to underscore that language use, both bilingual and monolingual, consists not only of the so-called core elements of language (phonetic and phonological knowledge, plus knowledge of morphology, syntax, and semantics) but also of the knowledge of the lexicon and the use of various registers. Hence, both monolingual adults and adulthood bilinguals can be claimed to be "victims" of imperfect acquisition, if their register repertoires and accompanying lexicons are restricted. Register expansion is potentially a life-long process, and register expansion brings together with it an expanding lexicon. We continue learning new words of our first languages (and languages we have subsequently acquired) if we have use for those words or interest in learning them. With reference to first language acquisition, Andersen (1990) remarks that the child's language experience is reflected in the range of registers the child learns to master. The wider a monolingual person's language experience is (i.e., the more speech situations the person has been exposed to), the more varied that person's command of registers (see also Finegan & Biber, 1994). The range of registers in which a person is competent does not necessarily have to correlate with education, since there are many academics who can be more or less at loss of words at a car repair shop, a doctor's office, the fish market, a computer hardware store, or a baseball game--all of course depending on their interests and disinterests. Some people are not at ease speaking with small children, since they are not used to those situations, and some avoid cocktail parties, not having ever acquired the smooth but often empty small talk that allows for seamless transitions from one chatting group to another. Therefore, the notion of imperfect acquisition, in this sense of the word, concerns us all. Lack of styles characteristic of certain registers cannot be considered attrition. In the same way, when looking at childhood bilingualism, attrition needs to be kept separate from imperfect acquisition.

Depending on the age when the L2 is introduced and the ties to comprehensive and rich input in the L1 (such as available in monolingual L1 environments) are severed, the effects of the curtailed input will likely manifest themselves in different areas of the L1. 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. But what has not ever been acquired cannot be lost: therefore, in early childhood bilingualism it is more appropriate to talk about incomplete acquisition of the L1.

In the case of my two research participants, S1 and S2, with whose language this paper is concerned, it is critical to determine to what extent they had acquired their L1 at the onset of their L2 acquisition (Seliger, 1996, pp. 607-608). Their acquisition of English began at the ages of six (for S1) and seven (for S2), when the girls immigrated into the United States from Finland. By that time, S2 had completed the first grade in Finland, and S1 had attended the Finnish kindergarten for three years. The girls were not tested formally for the level of their Finnish development; however, there is negative evidence available to allow the claim that their L1 had developed according to their age-level. The health-care system in Finland requires the children to be regularly examined at the local health-care center. During these visits, children are screened for all health and developmental problems: their vision and hearing are checked regularly, and part of the overall assessment is the informal evaluation of their language development by a trained pediatric nurse. Should any problems occur, the child will be sent to either psychological evaluations, speech therapy, or both. The subjects' health-care records indicate no irregularities, the only official note in the records being "clear speech" for S1 at the age of 3;3. There were no reports from the girls' kindergarten and first-grade teachers indicating delays in their language development. The informal records kept by the author (recordings and notes of the girls' language) indicate that by the age of three, S2 had acquired all the sounds of Finnish. At the same age, S1 was not yet producing the alveolar trill /r/, but she started to produce it at the age of 3;1. Their phonological development by the time they arrived in the United States was certainly completed.

Toivainen (1993) has studied the morphological development of 15 Finnish-speaking children between ages 1;0 to 3;5. According to Toivainen (1993, p. 303), by the age of 3;5, the average Finnish child has started to use the following morphological categories for the verb: imperative, negation, conditional, first and second person singular verb forms, second person plural, passive, past tense, negated past, perfect, past perfect, and first infinitive. For the noun, the child is already frequently using the plural and the following cases, in addition to the nominative case: inessive, elative, illative, adessive, allative, partitive, accusative, and genitive. Ablative and essive occur in the speech of some 3;5-year-olds. Thus, the most common ten Finnish nominal case inflections are already being used by the age of 3:5. At this age, the child is also starting to use comparative suffixes.

A comparison between the language of the children Toivainen studied and early notes and recordings of S1 and S2's language indicates that my subjects' acquisition of Finnish morphology by the time they were three years old was at the same level as that of Toivainen's research participants. The following excerpts, while simple in content and syntax, show already a range of verbal and nominal suffixes, as well as discourse particles. Excerpt (1) is a "bedtime story" invented ex tempore by S1 and dictated to the author when S1 was 2;11. S1 uses the illative, accusative, and adessive case inflections, the past tense, and is particularly fond of discourse particles:

 (1)   S1 (2;11) (Date: 9/9/85):  
    1 Milloin-ka-s                Kissa men-i metsa-an.  when-[Q.sup.3]-DP  
          cat go-PST  woods-ILL       se ol-i-kaan?       it  be-PST-DP  
      'When was it again?         The cat went to the  woods.'     2 Ott-i-kin       take-PST-DP       kissa kassi-n.  cat   bag-ACC  
      'The cat (went and)       took a bag.'     3  Tytto-kissa  
            Ja isa-kissa   ja  Jaakko-kissa.       girl-cat  
            And father-cat and Jaakko-cat       se ol-i.       it  be-PST  
      'It was the girl-cat.       And father-cat and  Jaakko-cat.'  
   4 Ja minu-lla ol-i-kin       And I-ADE   have-PST-DP       toinen  kissa.       other  cat       'And I had another cat.'     5  Ja toinen laht-i-kin        Se muist-i       ja laht-i.       and other  leave-PST-DP  
     it remember-PST  and leave-PST.       karku-un.  
    Metsa-an.       flight-ILL …