24Attention,+Memory,+and+the+_Noticing_+Hypothesis+

Language Learning 452 , June 1995, pp. 283-331 Review Article Attention, Memory, and the “Noticing” Hypothesis Peter Robinson University of Queensland Schmidt (1990) claimed that consciousness, in the sense of awareness of the form of input at the level of “noticing“, is necessary to subsequent second language acquisition (SLA). This claim runs counter to Krashen’s (1981) dual- system hypothesis that SLAlargely results from an uncon- scious “acquisition” system, the contribution of the con- scious “learning” system t o SLA being limited and periph- eral. Important to a theory of SLA that allows a central role to the act of noticing is a specification of the nature of the attentional mechanisms involved, and OftheirreIationship to current models of the organization of memory. With this in mind the present paper reviews current research into the nature of attention and memory and proposes a model of the relationship between them during SLA that, i t is argued, is complementary to Schmidt’s noticing hypoth- esis and oppositional to the dual-system hypothesis of Krashen. In light of this model, I argue that differential performance on implicit and explicit learning and memory experiments is caused by differences in the consciously regulated processing demands of training tasks and not by the activation of consciously and unconsciously accessed Correspondence concerning this article should be addressed to the author at University of Queensland, Centre for Language Teaching and Research, Brisbane, QLD 4072 Australia. Internet: peBrrBlingua.cltr.uq.oz.au 283 284 La nguage Learning Vol. 45, No. 2 systems. I also argue that the attentional demands of pedagogical tasks and individual differences in memory and attentional capacity both affect the extent of noticing, thereby directly influencing SLA. Schmidt (1990,1993a, 199333,1994a; Schmidt & Frota, 1986) has recently proposed that noticing, or conscious attention to the form of input, is necessary to subsequent second language (L2) development. A number of other researchers have also claimed an important role for “consciousness raising” activities and a role for “focus on form” in promoting L2 development (e.g., Fotos & Ellis, 1991; Long, 1988,1991; Rutherford, 1987; Sharwood Smith, 1991, 1993). These proposals run counter to claims that second language acquisition (SLA) is a largely subconscious process in which conscious learning serves merely t o monitor or edit an nonconsciously acquired knowledge base (Krashen, 1981, 1982, 19851, and that separate consciously and nonconsciously accessed systems of memory are differentially responsible for L2 learning processes (Paradis, 1994). Both positions have influenced L2 pedagogy. The task-based proposals of, for example, Nunan (19891, Long and Crookes (1992) and R. Ellis (19931, stress the importance ofnoticing and attention to form, though these propos- als differ with respect to methodological questions regarding how noticing should be facilitated and content questions regarding which aspects of language are important for learners t o notice. On the other hand, the attribution of SLA to unconscious processes has led t o the development of “The Natural Approach” (Krashen & Terrell, 19831, and has influenced the thinking underlying the Bangalore Procedural Syllabus (Prabhu, 1987, pp. 69-70). The issue of the role of conscious awareness in language learning is thus one where “deep divisions exist among language theorists” (Carr & Curran, 1994, p. 2151, and which is of fundamental pedagogic concern (Schmidt, in press). This paper aims to review research about two key cognitive mechanisms implicated in pro- posals for SLA based on the noticing hypothesis-attention and memory-and t o present a model of the relationship between Robinson 285 them, which I take t o be complementary to the noticing hypothesis. In the process I also identify other models that are oppositional (Beretta, 1991) in the sense that they could be taken t o provide theoretical support for Krashen’s dual-system hypothesis. By considering positions on the nature of attention and memory that support or contradict the noticing hypothesis, I hope to relate the debate about the role of consciousness in SLA t o relevant empirical work and operational constructs in cognitive psychology, and to thereby provide an extended theoretical basis for the noticing hypothesis. In the first section of this paper, I review the role of noticing in filter theories and capacity theories of attention and relate noticing to three functional classifications of attentional resources; alerting, orienting, and detecting. This review elabo- rates on a recent discussion of the role of attention during SLA (Tomlin & Villa, 19941, relating the concepts discussed t o the noticing hypothesis and t o a complementary model of short-term memory (Cowan, 1993). Although some discussion in the SLA literature has begun t o draw on cognitive models of attention in theorizing noticing during L2 development, there has been little discussion of the role of memory in noticing. With this in mind, in the paper’s second section I review various type classifications of memory-episodic versus semantic, procedural versus declara- tive, short-term versus Zong-term-considering whether these are indeed different systems, o r simply functional distinctions, and relating them to the discussion of attention. Resources, Information Processing, and Consciousness Schmidt’s (1990, 1993a, 199313, 1994a; Schmidt & Frota, 1986) claim that awareness at the level of noticing is necessary for converting L2 input t o intake invokes, but does not explain in detail, attentional mechanisms and their relationship to encoding and retrieval from the various subsystems of memory. Neither are attentional mechanisms and information-processing relationships explained in detail in other proposals which, following Schmidt, have attributed an important role t o noticing in L2 pedagogy (e.g., 286 Language Learn ing Vol. 45, No. 2 R. Ellis, 1993; Fotos, 1993; Fotos & Ellis, 1991; Long, 1991; Zalewski, 1993). Although there have been attempts to describe the relationship between input and intake at the neurophysiologi- cal level (Sato & Jacobs, 19921, there has been no attempt at a systems-level characterization of why attention is allocated to input under certain task conditions and not others. Attentional mechanisms, in Schmidt’s view, are causal factors in L2 learning, because they are responsible for allocating the cognitive resources that lead t o noticing, and subsequent encoding in memory. There are, though, competing characterizations of the role of attention during information processing; one recent explanation of implicit learning even claimed that the learning effects observed during a sequence learning task are due to a “nonattentional” form of processing (Curran & Keele, 1993). Part of the difficulty in motivating an explanation of such mechanisms using current cognitive theory lies in the fact, as Baddeley (1986, p. 225) noted, that the study of attention has been dominated by theories of the role of attention in perception and visual processes, particularly signal detection and pattern recog- nition. The role of attention in the control of memory and action, arguably areas of greater potential application to SLA processes, has been less well studied until recently. Recent work on the memory/attention interface has progressed considerably beyond the early multistore model of memory and attentional control of Atkinson and Shiffrin (19681, rejecting many of its fundamental assumptions (McClelland & Rumelhart, 1985). However, as Cowan (1988) noted, cognitive psychology has yet to settle on an accepted view of the mutual constraints imposed by memory and attention during information processing. Two established frameworks for describing skill development and performance, Shiffrin and Schneider’s (1977) theory of automaticity and Anderson’s 11983) ACT* theory of skill acquisition, have been heavily cited in the SLA literature (e.g., R. Ellis, 1993; Faxch & Kasper, 1984; Hulstijn, 1990; Kohonen, 1992; McLaughlin, 1987; O’Malley & Chamot, 1990; Robinson, 1989; Schmidt, 1992). However, more recent work in the study of action (Holding, 1989; Navon, 1984; Navon & Robinson 287 Gopher, 1980; Schneider & Detweiler, 1988; Wickens, 1980,1984, 1989), which describes the role of attentional processes in skilled performance using the dual-task paradigm for examining attentional allocation, has rarely been invoked by SLA research- ers, despite its potential relevance to such current issues as task complexity and grading in task-based approaches to L2 syllabus design (Long & Crookes, 1992; Nunan, 1988, 1989; Robinson, 1995a, t o appear). Part of the difficulty in motivating a theory of attentional mechanisms in SLA by drawing upon an accepted body of relevant findings from cognitive psychology research lies also in challenges t o traditional information-processing accounts of attention posed by more recent connectionist accounts. Information-processing models, such as Broadbent’s (1958), view attention as an executive process directing the serial passage of information between sepa- rate short-term and long-term memory stores. In contrast, connectionist accounts dispute the modular metaphor for cogni- tive architecture that the information-processing views are based on, as well as the assumption of seriality, arguing that executive attentional control is distributed throughout the entire processing system, in local patterns of neuronal excitation and inhibition, rather than in a central executive processor. Recent attempts t o reconcile connectionist and control architectures in the study of attention (e.g., Schneider, 1993; Schneider & Detweiler, 1988) are not yet widely accepted. In the following section I survey theories of attention and the functions of attention, and propose a model of the relationship between attention, awareness, and detection that is compatible with Schmidt’s (1990,1993a, 1993b, 1994a; Schmidt & Frota, 1986) noticing hypothesis. Attention The concept of attention has three uses. It can be used to describe the processes involved in “selecting” the information t o be processed and stored in memory. For example, researchers have used dichotic listening tasks to examine the fact that attention has 288 Language Learning Vol. 45, No. 2 a variable focus and can select information t o be processed t o the exclusion of other information (Cherry, 1953; Moray, 1959). It can be used t o describe our “capacity” for processing information. Studies of divided attention show that attention is capacity- limited and that decrements in performance increase as the number of task dimensions, o r components to be processed, in- crease (Taylor, Lindsay, & Forbes, 1967). Finally, it can be used to describe the mental “effort” involved in processing information. Pupillary dilation, for example, can be measured as a physical index of the degree of mental effort required in attending to increasingly complex tasks (Kahneman, 1973). Each of these uses has influenced the development of theories of attention. Filter Theories of Attention Early filter theories of attention were based on pipeline models of information processing, in which information is con- veyed in a fixed serial order from one storage structure to the next. In Broadbent’s (1958) “bottleneck” model, voluntary control of information processing is exercised by a selective attention mecha- nism or filter that selects information from a sensory register and relays it t o a detection device. Once past the selective filter, information is analyzed for meaning rather than for physical properties alone, enters awareness, and is encoded in short-term memory (Fig. 1). Such a model, based largely on acoustic process- ing, suggests that selective attention operates on the form of the message first and cannot simultaneously be directed a t form and meaning.’ If a mechanism of selective attention a t the stage of detection is responsible for noticing, then this model does not fit well with recent claims that manipulations in instructional treat- ments can encourage noticing via a focus on form, meaning, or a combination of both (Hulstijn, 1989; VanPatten, 1990) because all noticing would initially be form-focused. However, there are arguments against identifying noticing with the process of detec- tion (Fig. 4 and the discussion below) and subsequent develop- ments in filter theory disputed Broadbent’s claims about the Rob inson 289 Broadbent Treisman Sensory Selective Detection Short-term register filter device memory Sensor, Attenuation Delection Short-term register control devico memory R E S - 0 N S E = p --- - - - - - - - R E S E Figure 1. Selection in three models of attention and sensory process- ing. Note: From Cognitive Psychology (p. 43), by J. B. Best, 1992, New York: West Publishing. Copyright 1992 by Psychonomic Society Publications. Reprinted with permission. nature of detection. Treisman (19641, using evidence that partici- pants noticed their own name when repeated in an unshadowed ear during dichotic listening tasks, argued for a filter mechanism that was sensitive to semantic information as well as sensory information, and argued against the proposal that all information in an unselected channel is completely tuned out, and unavailable for detection. Arguments against Treisman's (1964) attenuatedfilter model are that the preattentive processing, or analysis before detection, it requires is too complete and resource-demanding. Late selection theories (Norman, 1968; Watanabe, 1980) have proposed that all information is processed in parallel and enters working memory, 290 Language Learning Vol. 45, No. 2 where a decision is made about its importance. Information judged important is elaborated or rehearsed; thatjudgedunimpor- tant is forgotten. Capacity Theories of Attention Underlying filter theories of attention and their associated mechanisms of selective attention is the metaphor of a limited- capacity channel, in which information competes for limited attentional resources available to the passive processor. More recent theories emphasize the voluntariness of the participant’s control of attentional resources and the task-specificity of deci- sions about attentional allocation. The metaphor most suited to these theories is that of attention as a spotlight, with a variable focus, which can be narrowed and intensified, or broadened and dissipated, as task conditions demand. Kahneman’s (1973) model allocated resources t o incoming stimuli from a pool of cognitive resources that varies as a function of the participant’s state of arousal. Allocation is divided between enduring predispositions (e.g., to recognize one’s own name) and momentary intentions (e.g., to eavesdrop). Divided attention does not necessarily lead to decrements in performance, given sufEcient arousal and given that the demands of the tasks performed concurrently are not excessive (Fig. 2). In this respect, capacity theories such as Kahneman’s differ from filter theories, which characterize incom- ing stimuli as inevitably involved in competition for limited resources. Wickens (1980, 1984, 1989) has recently expanded Kahneman’s (1973) view of attentional resource allocation, argu- ing that rather than a single pool of resources there are multiple pools. These pools occupy different points on three intersecting dimensions of resource systems: (a) the dimension representing perceptualhognitive activities versus response processes; (b) the dimension representing processing codes required by analog/ spatial activities versus verbal linguistic activities; and (c) the dimension representing processing modalities, that is, auditory Robinson Arousal -manifestations of arousal 291 Figure 2. Kahneman’s Capacity Model of Attention. Note: From Attention andEffort (p. lo), by D. Kahneman, 1973, Englewood Cliffs, NJ: Prentice-Hall. Copyright 1973 by Prentice-Hall, Inc. Reprinted with permission. versus visual perception and vocal versus manual responses (Fig. 3). Wickens (1989) argued that the attentional demands of tasks, and so their relative difficulty, will be increased when concurrently performed tasks draw simultaneously on the same pool of re- sources. In the worst cases, like maintaining two separate conversations, interference effects may make time-sharing impos- sible and the person adopts the attentional mechanism of “serial processing”-separate task components are completed in succes- sion. In other cases, where there is less “global similarity” between 292 Language Learning Vol. 45, No. 2 4 STAGES ----- w Cenlrol Encodhg Processing Rcspondnp Figure 3. Wickens’model ofthe structure of multiple resources. Note: From Human Skills (p. 82>, edited by D. H. Holding, 1989, New York: John Wiley. Copyright 1989 by Academic Press. Reprinted with permission. the tasks, and so less resource competition, as in driving a car while talking, the person may adopt the mechanism of parallel processing. However, although this may avoid the need for time- swapping, degradations in the quality ofthe attention allocated t o both activities may lead t o poor performance. When tasks draw on completely different pools of resources, or when one of the tasks is automatized, then successful time-sharing and dual-task perfor- mance are possible. In these circumstances “parallel processing” is always applied. However, Wickens noted, individuals may differ both in their time-sharing ability and in their store of available resources. Therefore, individual differences as well as task char- acteristics may determine which of these two mechanisms of attentional allocation the person adopts. Noticing, Attentional Theory, and L2 Task Demands Capacity theories, like those ofKahneman (1973) and Wickens Robinson 293 (1980, 1984, 19891, are properly seen as extensions of the late selection filter theory. In addition t o mechanisms of selective attention, they propose mechanisms for adjusting the deployment of attentional resources t o suit the particular conditions of task demands. In this sense, more recent attentional theory provides a framework for relating the act of noticing to those L2 task conditions that facilitate it. Recent task-based approaches to L2 pedagogy present learners with communicative activities in which attention is directed to meaning and t o the accomplishment of task goals requiring information exchange and joint problem solving. These activities contrast with form-focused approaches t o peda- gogy that explicitly direct learners’ attention to the formal properties of different types of input, such as formal grammatical rules, rehearsal of spoken dialogues, and so forth (Long & Crookes, 1992). Task-based approaches assume that noticing formal as- pects of language will be determined jointly by the learners’ stage of development and the structure of the task (R. Ellis, 1993; Long, 1989). Recent research has explored the extent to which design features of tasks can manipulate learner attention and cause incidental noticing of certain formal features of task input that learners may overlook if exposed to the L2 in untutored conversa- tional settings (Doughty, 1991; Hulstijn, 1989; Watanabe, 1992). One of the main claims for the advantages of task-based instruc- tion therefore, is that well-designed tasks can facilitate noticing of aspects of L2 synta