REVIEW OF LITERATURE
This chapter comprises six main sections that summarize the theoretical and empirical knowledge base regarding second language development and academic achievement in a second language. At the end of each main section,4 there is a summary synthesizing studies and highlighting key findings relevant to the present study. The first section reviews selected second language acquisition theories that reflect representative, current trends in the field and provide a theoretical foundation for the study. The second section focuses on defining language proficiency and reviews relevant studies illuminating linguistic factors implicated in ELLs' schooling. The third section summarizes the language learning strategy research base. The fourth section identifies relevant theories and research regarding motivation for learning a second language and makes connections among motivation, language learning strategies, and proficiency. The fifth section reviews studies that have examined academic achievement in second language. The last section summarizes the present study's variables as generated from this review of the literature.
Second Language Acquisition
Defining the Field
Second language acquisition (SLA) is an interdisciplinary field that is both historically old and new (Gass & Seliker, 2008). The field is old because the nature of 4 Except the first section which integrates theory and empirical findings throughout. 16
second language learning and teaching has fascinated scholars for centuries. In more modem terms, SLA is a young discipline which, beginning in the 1960s, distinguished itself from applied linguistics and education (Long, 2006). SLA focuses on second language (L2), as well as second dialect, learning and loss by children and adults. As formulated by Saville-Troike (2006), SLA seeks to answer three main questions: ( a) "What exactly does the L2 learner know?" (b) "How does the learner acquire this knowledge?" and (c) "Why are some learners more successful than others?" (p. 24). From its beginnings, SLA has taken a multidisciplinary approach and drawn on other established disciplines including education, linguistics, psychology, anthropology, and sociology (Gass & Seliker, 2008).
The theoretical rationales in considering the connections between SLA and academic achievement research are twofold. First, as discussed in Chapter 1, a current focus in educational literature is on linguistic factors implicated in ELLs' schooling. In fact, many scholars find it difficult to distinguish between language proficiency and academic competence (Abedi & Lord, 2001; Solorzano, 2008); some scholars speak in terms of academic English development (SchleppegreU, 2004) or of academic achievement in second language (Collier, 1987; Collier & Thomas, 1989). Second, language learning strategies, the primary variable of interest to this study, have been chiefly examined from the SLA perspective.
Selected SLA Theoretical Frameworks
Language as a faculty of the mind. Early research on second language (L2) development was strongly influenced by research on first language (L 1) acquisition and by nativist linguistic theories. Nativist theories posit that language acquisition is 17
accomplished through the use of innate linguistic abilities. Many nativists assert that L 1 and L2 acquisition are similar in that they rely on essentially the same processes. Chomsky (1959) hypothesized that language learners construct a theory of grammar from linguistic input using a "built-in," genetically endowed hypothesisforming device, also known as the language faculty or the language acquisition device (LAD). Contrary to then-popular behaviorist ideas (Skinner, 1957), Chomsky argued that language acquisition cannot be solely attributed to learning through stimulus-responsereinforcement mechanisms. First, he pointed out that proficient speakers of a language can produce and comprehend novel, unheard-of sentences. Rather than a learned repertoire of prefabricated responses, as proposed by behaviorists, this creative ability suggests the existence of a special program, or language faculty, that assists humans in generating "an unlimited set of sentences out ofa finite list of words" (Pinker, 2007, p. 9).
Second, Chomsky noted the fact that all children, largely independently of intelligence, construct grammar in comparable ways within a remarkably short time and often without explicit reinforcement on the part of their parents (as in the case of young English language learners). Chomsky proposed that these two phenomena could be explained by speakers' individual contributions such as "inborn structure, the genetically determined course of maturation and past experience" (p. 27). Pinker (2007) termed this inborn structure or program language instinct and argued that language was "a biological adaptation to communicative information" (p. 5). Chomskian ideas received a large following among linguists and psychologists; as a consequence, language acquisition came to be understood as a gradual, largely unconscious process of syntactic rules 18
deduction and subsequent acquisition. Rules deduction was thought to be triggered by linguistic input and assisted by LAD.
Nativist views on L2 acquisition are exemplified by Krashen's theory ofSLA. Krashen (1985, 1987) argued that an L2 was acquired, for the most part, unconsciously and in the presence of what he termed comprehensible input (i.e., input that contains linguistic features slightly beyond the leamer's current level of proficiency). One of the key hypotheses of the theory, the Acquisition/Learning hypothesis, distinguished between language learning and language acquisition. Language acquisition was said to be an unconscious process of rules extraction equally accessible to children and adults. Language learning, on the other hand, was defined as the conscious learning of rules through formal instruction and error correction.
An additional component ofKrashen's theory, the Affective Filter hypothesis, was proposed to account for individual differences in language acquisition. According to this hypothesis, affective factors (e.g., anxiety, motivation, or self-confidence) influence, on subconscious level, how much information L2 learners extract from available input. Krashen (1985) argued that the Affective Filter hypothesis was supported by research documenting faster development in "lower filter" versus "higher filter" L2 learners exposed to the same amount ofL2 input. In sum, according to Krashen's SLA theory, people acquire an L2 when two conditions are met: (a) input is made available and comprehensible for language learners, and (b) learner Affective Filter is low to allow the intake of the input.
While Krashen's SLA theory attracted many followers and had a tremendous impact on L2 pedagogy in the 1980s (Spada, 1997), classroom research has provided 19
counterevidence to Krashen's hypothesis that speaking would naturally "emerge" in the presence of comprehensible input. Studies have shown that in input-based instructional programs (such as L2 immersion), L2 learners achieved native-like levels in comprehension (reading and oral comprehension), but not in production (writing and speaking; Swain, 2005). Moreover, contrary to Krashen's predictions, explicit instruction was found to benefit language learning. In her review of over 40 descriptive, quasiexperimental, and laboratory studies conducted with school-aged and adult L2 learners, Spada concluded that, when combined with a focus on meaning, form-focused instruction (i.e., instruction that deliberately focuses learners' attention on L2 grammatical features) does make a positive difference in L2 acquisition.
Moreover, many researchers found Krashen's SLA theory to be inadequate when it came to explaining a much greater variability in older L2 learners (Gregg, 1984; Robinson, 1995; Schmidt, 1990). Gregg, for example, argued that older learners relied on their superior cognitive (not language-specific) abilities oflogic and problem-solving in order to "construct" L2 grammar rules. Additionally, the fact that only in rare cases did older learners achieve native-like proficiency (Bialystok & Hakuta, 1999; Hakuta, Bialystok, & Wiley, 2003; Long, 1990) led some SLA researchers to conclude that Ll and L2 acquisition differed in some fundamental ways. Newport (1990), for example, proposed that older learners' superior cognitive abilities of memory and perception (although associated with acquisition of larger chucks of language) prohibited componential analysis available to young children, thus accounting for the child-adult discrepancies in ultimate L2 attainment.
Several authors observed that the role of consciousness in L2 learning was considerably greater than suggested by some Ll theories (Robinson, 1995; Schmidt, 1990). O'Malley, Chamot, and Walker (1987), for example, argued that cognitive learning theories, discussed next, may provide a more appropriate framework for understanding and studying processes involved in L2 acquisition. Language as a cognitive skill. Whereas nativist linguistic theories hold that language is a genetically endowed faculty of the mind, learning theories (Anderson, 1982, 1989; Skehan, 1998) posit that L2 is a complex cognitive skill whose acquisition is comparable to the learning of other complex skills such as problem solving. That is, linguistic codes are thought to be acquired, stored, and retrieved from memory much in the same way as any other information (O'Malley et aI., 1987). Learning is believed to be the result of language processing itself, where linguistic knowledge ("knowledge that") is transformed into linguistic performance ("knowledge how") through rules extraction, learning, and automatization (Long, 2006; Saville-Troike, 2006). Cognitive learning theories thus study how new information, including new L2 information, is processed and stored in memory. One of the cognitive theories that informed SLA research, and in particular LLS research, has been Anderson's Adoptive Control of Thought (ACT) Theory (Chamot, 2005b; O'Malley et aI., 1987).
Anderson (1982, 1989) distinguished between two types of knowledge (or memory): (a) declarative knowledge, information stored in the form of facts; and (b) procedural knowledge, production rules mentally represented in the form of IF ITHEN5 (condition-action) pairs. This declarative-procedural knowledge distinction roughly 5 One example of an IF ITHEN pair is, "IF the goal is to generate the present tense of HUG, THEN say HUG + s" (Anderson, 1989, p. 325).
corresponds to what Chomsky termed the distinction between linguistic competence (i.e., mentally represented knowledge of grammar) and linguistic performance (i.e., the use of linguistic knowledge in language comprehension and production; as cited in Fromkin, Rodman, & Hyams, 2007).
According to Anderson's (1989) ACT model of information flow, the interplay between declarative and procedural knowledge, which takes place in working memory, results in the learning of complex cognitive skills. Figure I graphically represents the flow of information within the ACT framework.
RETRIEVE MATCH &
Figure 1.1. The flow of information within the ACT framework (Anderson, 1989, p. 319).
Anderson (1982) proposed three stages6 of the skill acquisition process: (a) declarative, (b) knowledge compilation, and (c) procedural. In the declarative stage of skill acquisition, the learner consciously extracts, or receives through instruction, factual information (e.g., patterns or rules) about a skill. This declarative knowledge is rehearsed and interpreted in short-term memory to generate "primitive rules" (productions) that specify the type of actions to be taken under a given set of circumstances. At this stage, a leamer's behavior is guided by a trial-and-error exploration in which the already existing productions (stored in long-term, declarative memory) are applied to new information. This interpretive process is largely conscious and places a rather taxing demand on the leamer's processing capacities. O'Malley et al. (1987) argued that in terms ofL2 developmental stages, Anderson's declarative stage may explain the silent period-also known as delayed production-a period during which language learners do not engage in L2 conversations. During this time, L2 learners are focused on developing a knowledge base about the new language. When first attempts at communication are made, beginning-level language learners with limited L2 declarative knowledge often recur to their L 1 linguistic forms (e.g., vocabulary, grammatical structures, and discourse patterns).
The next stage, knowledge compilation, has been described as a "gradual process by which the knowledge is converted from declarative to procedural form" (Anderson, 1982, p. 370). This is realized by two processes: (a) collapsing simple production rules into single rules with an effect of a sequence (metaproductions); and (b) eliminating references to the declarative knowledge (proceduralization). That is, to perform a task, 6 In some sources, these three stages are referred to as cognition, association, and autonomy (McDonough, 1999).
the learner no longer needs to retrieve declarative knowledge into working memory. This stage results in a considerable speedup of the processing and ultimately allows the learner to complete a task in a single trial without verbal rehearsal in short-term memory. According to O'Malley and colleagues (1987), this stage may correspond to the development of what linguists termed interlanguage (i.e., learner intermediate linguistic system between an Ll and an L2).
In the procedural stage, the learner becomes more efficient through experience (Anderson, 1982, 1989). This process involves the gradual fine-tuning of metaproductions; that is, general rules transferrable to novel situations are extracted and the conditions under which these rules can be applied are specified. During this stage, the amount of practice and, by implication, the ability to generate opportunities for practice are key to successful skills development. This is because every time that factual or procedural information is "fired" (selected) for the performance of a particular task, knowledge is strengthened. In SLA, this stage is described as automatization (i.e., the development of effortless and largely unconscious skill performance; Skehan, 1998). Automatization corresponds to fluency development, in which the performance of a language learner gradually approaches that ofa native speaker. O'Malley et al. (1987) argued that while declarative knowledge can be acquired relatively quickly, the development of procedural knowledge requires a prolonged period of time. The authors also noted that declarative knowledge is not a sufficient condition for L2 production. An alternative explanation of processes involved in language learning was proposed by social theories. These theories "underscore the social nature" of both Ll and 24
L2 learning (Gersten & Hudelson, 2005, p. 23). The next section discusses one such theory, namely Vygotsky's (1978) sociocultural theory.
Language as a socially constructed skill. Unlike cognitive learning theories, social theories consider the quality of contact with new language (i.e., social interactions), and not the amount of L2 experience to be at the core of L2 acquisition (Long, 2006). Children are believed to develop language and, importantly, cognition over time through multiple interactions with more capable others (adults and peers). According to Vygotsky's (1978) sociocultural theory, the nature of social interactions defines the what (the content) and the how (how information is interpreted, organized, and retained) of the child's learning. Learning occurs through the transformation of simple mental activities (e.g., labeling) to higher mental activities (e.g., abstraction) by means of language, mediation, and internalization (Leong & Bodrova, 1995). Thus, the development of language and cognition are perceived as interdependent processes. Vygotsky (1978) wrote, "the only 'good learning' is that which is in advance of development" (p. 89). Such "good learning" can only happen with expert assistance and within the Zone of Proximal Development (ZPD), which he defined as follows: "It is the distance between the actual developmental level as determined by independent problem solving and the level of potential development as determined through problem solving under adult guidance or in collaboration with more capable peers" (p. 86). In early developmental stages, language is external to the learner. First, children use language to solve problems on an interpersonal level (within the ZPD) when they negotiate the task with more capable others (adults or more advanced peers). Children then internalize language to develop inner or intrapersonal speech: "Instead of appealing 25
to the adults, children appeal to themselves" (p. 27). Thus language becomes the planning device for organizing one's conceptual understandings with a sociocultural motivation to share with others.
As applied to L2 development, the concept of ZPD may be realized through what SLA researchers term modified input (i.e., the use of short, low syntactic complexity utterances) or modified interaction (i.e., frequent comprehension checks, clarification requests, and self- and other-repetition; Long, 1983, 1987). Modified input and interaction are the means by which language experts provide language learners with ready-to-use chunks of speech, thus enabling L2 learners to express themselves beyond their current means (Saville-Troike, 2006). Among intrapersonal speech types observed in L2 learners, Saville-Troike distinguished: ( a) private speech, an audible talk to self, usually present in young children; (b) private writing, writing for oneself (e.g., lists of new words, translations into L 1, or notes on the margins of a book); and (c) inner speech, inside-the-head speech. Saville-Troike reported on studies documenting L2 learners using intrapersonal speech to actively build up their L2 competence even when they were not directly engaged in L2 communication.
In sum, from Vygotsky's (1978) sociocultural perspective, learning is realized through interpersonal and intrapersonal social mediation. This model predicts differentiated performance among L2learners depending on: (a) learners' access to and amount of participation in a learning community; (b) the amount of mediation from experts and peers; and (c) the degree to which learners use that help (Saville-Troike, 26
2006). Similarly to Anderson's (1982, 1989) theoretical work, Vygotsky's theory has been informative to LLS research.7