Part 1
Multilingualism in Education: Conceptual Issues and Sociolinguistic Perspectives
1 Problems in Defining the Concepts of L1, L2 and L3
Bjƶrn Hammarberg
Introduction
With the increased interest in multilingualism and third language acquisition (TLA) in recent years, and the rapidly expanding research in the area, several authors have felt the need to review and reflect on parts of the terminology used in this field (Aronin et al., 2011; De Angelis, 2007; Franceschini, 2009; Hammarberg, 2010; Kemp, 2009). When the research perspective is shifted from a bilateral view, which has prevailed in language learning theory (bilingualism; L1/L2; second language acquisition) to situations that explicitly involve more complex language settings and new areas of inquiry, the use of established terms may in some cases lead to complications. This is true even of very basic terms, such as first language (L1), second language (L2) and, not least, third language (L3), which will be at the centre of the following discussion.
The use of the terms L1, L2 and L3 in the literature is quite disparate. Thus, for example, in the area of second language acquisition (SLA) research, the term āsecond languageā or āL2ā usually refers to (a) any non-native language in a speakerās repertoire, whereas in third language acquisition (TLA) studies, it is often used for (b) the chronologically second language acquired by a speaker. In the field of TLA, or language acquisition in a multilingual context, āthird languageā or āL3ā may stand for (a) the chronologically third language (a frequent conception), or (b) the next language encountered after the simultaneous acquisition of two languages in early infancy (Cenoz, 2000), or (c) any non-native language currently being acquired by a speaker who is already familiar with one or more other non-native languages (e.g. Williams & Hammarberg, 2009 [1998]). In addition (d), the notion āthird or additional languageā is used instead of āthird languageā (De Angelis, 2007).
The purpose of the following considerations is to try to elucidate this rather inconsistent usage of the term, identify its various defining criteria and discuss the basis for adopting a coherent and adequate terminology in this area of research. In the first place, this is a matter of identifying the concepts for which there is need and how these are related to each other, and in the second place to consider how they can best be named by terms. It appears timely to discuss this issue while TLA studies are still a fairly new branch of research, in the process of becoming more firmly established.
We can observe an increasing focus today on language acquisition in a multilingual context.1 It is being taken into account that peopleās acquisition and alternate use of several languages is a natural and extremely widespread phenomenon. Many researchers assume bi- or multilingualism to be as frequent or more frequent in the world than monolingualism (see e.g. Aronin & Singleton, 2008: 2, 2012: 41; Cook, 1992: 578; de Bot, 1992: 2; Grosjean, 1982: vii; Hakuta, 1986: 4ā6; Kecskes, 2010: 93; Tucker, 1998: 4). It is a fact that all humans have the capacity to acquire and use several languages. Speakers tend to develop bi- and multilingual competence in the course of their lives, in some cases starting in early infancy. It can be argued that the multilingual potential is an integral part of the human language faculty, and that multilingualism, rather than monolingualism, should be seen as the normal form of (mature) linguistic competence.
In this perspective, the developing multilingual competence is a key area of research. Some of the fundamental aspects of this competence are mentioned as follows: One is its holistic nature, the way a speakerās competence in various languages forms a unique and coherent whole, a multicompetence (Cook, 1991, 1992, 1993; cf. Grosjean, 1985), rather than constituting separated monolingual competences. The speakerās languages are assumed to form subsets within the same cognitive system (Paradis, 1981, 2004). They get simultaneously activated to varying degrees during comprehension and production (Green, 1986), which forms a basis for crosslinguistic influence (CLI). Depending on various factors in the speech situation, a speaker may adopt different language modes: a monolingual mode, striving to adhere to the use of one language, or a bi- or multilingual mode, allowing transfer and code-switching more freely (Grosjean, 2001).
Another aspect concerns the hierarchical relations between the languages in the speakerās mind, relations which have motivated the use of rank designations such as primary versus secondary system (Weinreich, 1953: 14), tertiary language (Hufeisen & Lindemann, 1998) or the common ordering of speakersā languages as first, second, third and so on (L1, L2, L3ā¦). A crucial question then is, what is the nature of this hierarchy? It is essential to clarify this question in order to understand the concepts that are commonly called L1, L2 and L3.
The following section will deal with identifying the hierarchical relations between a learnerās languages on which current research in the area is based, starting with the basic distinction between a native and a non-native language, and continuing with more complex situations involving more languages. This will result in identifying several alternative models that underlie current conceptions of an acquisitional hierarchy. In the subsequent section, the properties and problems connected with these models will be discussed, as well as implications for the use of terms. The concluding section will comment on the rise of a terminology which is no longer quite adequate, and ways of handling the current terminological situation.
Identifying a Hierarchy for the Acquisition of Languages
L1 and L2 in the SLA tradition
An individual acquires languages during different periods of life, be it simultaneously, successively or in reciprocal alternation. Various factors in the acquisitional history (age, chronology, amount of exposition and use, context and manner of acquisition, affective factors, etc.) contribute to differences in how the speaker processes, acquires and masters the different languages. In the SLA tradition, a simple division is made between L1 and L2. A basic question then is, what is the fundamental difference between an L1 and an L2?
Among the various aspects in which a personās languages may differ (such as time of acquisition, level of proficiency, amount and domains of use, subjective identification; cf. discussion in Skutnabb-Kangas, 1981: 12ā16), the type of criterion taken to define the concepts of L1 and L2 is usually one of time-relation. Since the alternative rank orderings of a personās languages according to time, proficiency, use or identification often do not yield the same result, it is important to keep these criteria apart so that it is possible to profile them against each other. The current discussion of criteria for defining L1 and L2 will thus focus on the time dimension. Two major time-related criteria can be discerned in the SLA literature:
(1) Chronological order of acquisition. This means that L1 and L2 are defined through the priority/posteriority distinction. L1, being encountered first, develops as the original system, and L2 is subsequently added to the already established L1.
(2) Cognitive maturity. This criterion is based on the cognitive development that comes with age during early childhood. The crucial distinction is the one between a language encountered and acquired up to a certain level during infancy, the period in which the childās linguistic categories, patterns and rules of use are first shaped, and a language encountered at a later age. This is not just a chronological division, but it reflects a difference in the individualās stages of maturational, social and intellectual development. The distinction is the one between a native language (NL) and a non-native language (NNL).
In a simple case with one language acquired from birth and another added after early childhood, the terms L1 and L2 can be interpreted according to either of the two criteria. But when these terms are also applied to more complex cases where a speaker has multiple NLs and/or NNLs, so that L1 comes to be used for one or more NLs and L2 for one or more NNLs, as has become customary in SLA research, this implies that the distinction is made according to the cognitive maturity criterion: any NL is designated as an L1 and any NNL as an L2. This has in effect become the standard conception of L1 and L2 in the field of SLA.
The validity of the NL/NNL distinction is evident when a language acquired from birth is compared to a language added in adolescence or adulthood. Thus, in contrast to an NL, we can observe that an NNL is usually subject to fossilisation, and that crosslinguistic influence tends to affect NNLs in regular ways, symptoms that bear witness to cognitively based differences between NL and NNL. In a neurolinguistic context, Paradis (2004, 2008, 2009) distinguishes between implicit linguistic competence in L1 and explicit metalinguistic knowledge of L2. His use of L1 and L2 here corresponds to NL and NNL. To summarise, the former is acquired incidentally, stored implicitly, used automatically, sustained by procedural memory, and involves different parts of the brain than the latter, which is learned consciously, stored explicitly, consciously controlled when used, and sustained by declarative memory (Paradis, 2008: 343). Paradis emphasises that the two mechanisms are distinct, but he allows for gradual replacement of explicit knowledge by implicit competence in L2 users through practice. āAs a skill becomes more proficient, processing shifts from the use of one mechanism (controlled, declarative) to another (automatic, procedural). [ā¦] Practice leads to the replacement of controlled processes by automatic processes, thus improving automaticityā (Paradis, 2004: 36, italics in original).
A process of shift of mechanisms also seems to take place when an infant grows older and the NL type of acquiring and storing a new language is gradually replaced by an NNL type. For a period, the young child is thus likely to undergo a transition process during which a new language shows both NL and NNL characteristics. Different researchers vary somewhat as to how this process should be located in time. Paradis (2008) sets the endpoint for L1-type acquisition at about age five. McLaughlin (1984) posits age three as a practical cutoff point between FLA and SLA. Meisel (2011), reviewing research on neural maturation and age, finds that changes in the direction from L1-type to L2-type acquisition seem to take place at least up to age seven, with some crucial changes around age four. Hyltenstam and Abrahamsson (2003), investigating age of onset of acquisition and ultimate attainment, report effects that appear to indicate that the childās capacity for L1-type acquisition starts decreasing successively already shortly after birth. What is constant in these findings is that the transition from NL to NNL acquisition proceeds with a temporal overlap, which, however, does not rule out the relevance of the NL/NNL distinction.
Beyond a two-level distinction
The exploration of TLA, or more generally, language acquisition by previously bi- or multilingual persons, means that the complexity of the learnerās language background is being taken into account. A basic finding is that being already bi- or multilingual creates a different basis for acquiring a further language as compared to acquiring an NNL for the first time. There is evidence for various positive effects of bilingualism on cognitive development, metalinguistic awareness and communicative skills, as well as on attaining proficiency in a further language; cf. Cenoz (2003) for a review of research. An open question which still needs closer study and clarification is whether it makes a difference in these respects if the speakerās biling...