Before the birth of the SLA field several decades ago, non-SLA fields had been using the term “learning,” and still do, to describe stimuli such as colors, num-bers, lights, dark spots on a computer, and so on. With the birth of SLA, research-ers have placed a premium interest in the exploration of the construct of learning as in second or foreign language (L2), learning in the naturalistic (immersion or study abroad), classroom, and laboratory settings. If we were to take a quick survey of any representative number of published theoretical or empirical studies in both SLA and non-SLA fields, we would find an inevitable mention of the term “learning.” At the same time, it is quite revealing that a closer and more careful look at what comprises “learning” within and between the SLA and non-SLA fields may not be the same construct. For example, the concept of intake is not well acknowledged in many non-SLA fields, and whatever is taken in may be viewed as learning. In addition, there may be quite a lot of terminological confusion given that the construct of learning appears to be operationalized or measured by quite a wide range of assessment tasks.
In an effort to avoid terminological confusion, this chapter proposes looking at the construct of learning from a tri-dimensional perspective, namely (1) learn-ing as a process versus learnlearn-ing as a product, (2) the kind of learnlearn-ing, namely item versus system learning, and (3) the type of processing involved, namely explicit (i.e., with awareness) versus implicit (i.e., without awareness). In addition, this construct will be situated within the stages postulated to occur along the learn-ing process in SLA ( Chapter 2 ). This global and theoretical view of the learnlearn-ing process has a three-fold purpose: It allows researchers to (1) identify which stage along the learning process learning is being investigated, (2) address the appropri-ate assumptions in relation to the tri-dimensional perspective of learning, and (3) interpret the results within this theoretical framework. A visual framework
for researchers to measure and interpret the process and product of learning will be provided, together with a discussion of both receptive and productive mea-sures employed to address learning. Finally, I shall administer a test to address this construct (so take good notes). But first, a terminological issue to discuss.
Acquisition Versus Learning
Before we begin to deconstruct the construct of learning, I would like to situate this construct in relation to the term “acquisition” and raise one basic question:
Is the learning process the same as the acquisitional process? It would appear that some researchers do not see any difference between these two processes, and this conf lation is observed in both SLA and non-SLA fields. In other words, there appears to be a somewhat lax approach in reporting these two processes in terms of what type of processing was specifically addressed. For example, two recent articles in cognitive science appeared to indicate that the two processes are similar:
“higher order concept learning ”
“. . . the other set was focusing on acquiring the individual input-output pairs (exemplars) vs. . . .”
“this set of participants was attempting to learn the function rule during training.”
(McDaniel, Cahill, Robbins, & Wiener, 2013: 18–19, emphasis added) “Chinese participants learned more knowledge . . . Chinese participants acquired greater knowledge . . .”
(Fu, Dienes, Shang, & Fu, 2013: 9, emphasis added) In SLA, similar conf lation is easily found in, for example, the title of God-froid, Boers, and Housen (2013), which included a reference to incidental L2 vocabulary acquisition while the authors discussed “word learning ” in their inter-pretation of the data. Stafford, Bowden, and Sanz (2012) reported that their interpretation test assessed “what they had learned during treatment” (p. 752) while their production test assessed “to what degree knowledge of Latin mor-phosyntax acquired through interaction with the input-based treatment could be transferred to a guided production task” (p. 753). Indeed, this conf lation may be due to the name of the field in which we conduct our research, namely, SLA.
As I mentioned in Chapter 1 , Krashen’s (1982) Monitor Theory was the first theoretical underpinning in those days to raise the issue of the role of the con-struct of awareness or “consciousness” in the L2 learning process and to dis-tinguish between learning (with consciousness), resulting in learned/explicit knowledge, and acquiring (without consciousness), resulting in acquired/implicit knowledge. Now we need to consider the source of Krashen’s model, namely
L1 acquisition and, more specifically, children’s L1 acquisition. Acquisition of a language takes place in the environment in which we learn our first language, surrounded by this language and used to communicate, read, share information, and so on. We acquire our L1 by living it, to put it mildly, with minimum effort on our part to acquire it. L1 processing is largely unconscious and takes place over a long period of time until we hit kindergarten (or beyond), where we begin to be exposed to explicit instruction, including how to read, write, pro-nounce words properly, and so on. In this formal environment, we begin to learn explicitly how to use our first language, and this continues for quite some time (it is all relative) after. The one environment that can offer an L2 learner a similar acquisitional condition is that of the immersion setting, with the understanding that the L2 learner will engage in similar activities as the L1 speaker did over a relatively long period of time. To this end, the learning process is distinct to the acquisition process due to obvious reasons and this distinction is followed in this book. Now let us begin our deconstruction.
The Construct of Learning
First, let me ask a very basic question (both researchers and teachers share the same love for asking questions): How do you define learning? Take a pause and think a bit deeply (put in some cognitive effort) about this question. When someone says that learning is taking place or took place, are these two statements referring to the same idea or are they conceptually different? Dictionary defini-tions of what comprises learning view this construct from minimally two main perspectives, namely as a process (verb) and as a product (noun). As a process, it is defined in Wikipedia as “acquiring new, or modifying existing knowledge, behaviors, skills, values, or preferences, and may involve synthesizing differ-ent types of information,” and in the Merriam-Webster dictionary as “to gain knowledge or understanding of or skill in by study, instruction, or experience.”
As a noun, it is the knowledge or skill acquired by instruction, study, or expe-rience (Merriam-Webster Online). Based on these definitions, the construct of learning appears to involve both old and new information with the potential for the old information (prior knowledge) to be modified based on the new infor-mation. However, note that these definitions do not delve explicitly into whether learning (as a process) (1) can be focused on discrete items in the input (item-learning), resulting in knowledge (as a product) comprising learners’ subsequent ability to recognize or even produce these individual items, or (2) involves some internalization of rules underlying such discrete items (system-learning) result-ing in knowledge (as a product) comprisresult-ing learners’ subsequent ability to ver-balize a grammatical rule or, minimally, to demonstrate an ability to generalize this underlying rule to new exemplars. OK, you may go back and re-read that last sentence (keep in mind the stages of the learning process), but I am going to elaborate on it later. In addition, these definitions do not offer much insight
into whether the process and product of learning involves a role for awareness (explicit learning and explicit knowledge, respectively) or lack thereof (implicit learning and implicit knowledge, respectively). Let us first take a look at some operationalizations of what comprises learning.
Operationalizing the Construct of Learning
A close review of many empirical studies in both SLA and non-SLA literatures will easily demonstrate that the construct of learning is rarely operationalized.
Here are two recent noteworthy attempts to provide an operationalization in both non-SLA and SLA fields, just to give you a feeling of different perceptions.
This recent operationalization (McDaniel et al., 2013) from cognitive psy-chology appears to posit levels of learning:
1. Learning: Learners were distinguished from non-learners as having a mean absolute error of less than 10 during training.
2. Exemplar learning: Learning of the individual cue-criterion pairing for the 20 training points was operationalized as showing relatively f lat extrapola-tion after having met a strict learning criterion (the one above).
3. Rule learning: Learning of the relations/rules among training points was operationalized as showing general extrapolation along the slopes of the bilinear function.
This recent operationalization (Stafford et al., 2012) from SLA appears to posit different stages (initial vs. late?) of learning in relation to specific off line assess-ment tasks:
Initial language learning was operationalized as follows: overall accuracy on tests of written interpretation, aural interpretation, grammaticality judgment, and written production.
(p. 754) The failure to operationalize or even to define the construct of learning in most studies has led one to seek this operationalization in either what (e.g., colors, visual stimuli) is being measured in non-SLA fields or the assessment measures employed in SLA to address this construct. The many measures of “learning”
appear to indicate several interpretations of what comprises the construct of learning, leading to an inevitable terminological confusion. For example, learn-ing has been measured by a series of receptive tests such as recall or rememberlearn-ing (e.g., Shekary & Tahririan, 2006), trials-to-criterion (the number of attempts it takes to start processing the input correctly; e.g., Fernández, 2008), selec-tion/recognition (Godfroid et al., 2013; Leow, 2000), selection/interpretation (Williams, 2005), a four-alternative, forced-choice picture matching task
(Hamrick & Rebuschat, 2012), productive tests such as written interpretation (Stafford et al., 2012), acceptability judgment (Grey, Williams, & Rebuschat, 2014), fill-in-the-blank (Medina, 2015), modified cloze test (e.g., Rossomondo, 2007), and so on. In addition, learning has also been measured concurrently via reaction time (Leung & Williams, 2014) and non-concurrently (most empiri-cal studies), etc. Embedded in these measures are several perspectives of what comprises learning, as evident in whether learning is viewed as a process (e.g., Rosa & Leow, 2004a) or a product (e.g., Williams, 2005), the kind of learning assumed, that is, item (e.g., Leow, 2000) versus system learning (e.g., Rosa &
Leow, 2004b), or the type of learning assumed, that is, implicit or explicit learn-ing (Chan & Leung, 2014).
A Tri-Dimensional Perspective of the Construct of Learning In an effort to avoid terminological confusion, I propose that we look at the con-struct of learning from a tri-dimensional perspective, namely, (1) learning as a process versus learning as a product, (2) the kind of learning, namely item versus system learning, and (3) the type of processing involved, namely explicit (i.e., with awareness) versus implicit (i.e., without awareness). In addition, we situate this construct within the stages postulated to occur along the learning process in SLA ( Chapter 2 ). This global and theoretical view of the learning process has a three-fold purpose: It allows researchers to (1) identify which stage along the learning process learning is being investigated, (2) address the appropriate assump-tions in relation to the tri-dimensional perspective of learning, and (3) interpret the results within this theoretical framework.
Learning as a Process Versus Learning as a Product
If you recall the postulated stages of the L2 learning process in Chapter 2 , you will recall that learning as a construct may be viewed from two perspectives:
(1) Learning as a process , which occurs internally, takes place at Stages 1 (input processing), 3 (intake processing), and 5 (L2 knowledge/output processing) and (2) learning as a product (what is learned) is presented internally (at Stage 4) in the learner system as knowledge, and externally as representative L2 knowledge.
Stage 2 represents intake as an initial product kept in working memory that may be retrieved immediately via concurrent receptive tests that only require learners to recognize, select, or identify discrete target items presented in the input, but has yet to be further processed, internalized, or learned.
In other words, when we report about the learning process , we are referring to the process of converting input into intake, the process of converting intake into the internal system, which is typically assumed to contain some type of knowl-edge (systemized or discrete/un-systemized items, explicit or implicit, declara-tive or procedural), and there is also the process of producing output that allows
the learner to potentially receive additional L2 input that allows either a confir-mation of his/her L2 knowledge or a restructuring of his/her interlanguage. To measure any stage of the learning process , concurrent data elicitation procedures that can provide insights into learners’ thought processes may be more appropri-ate than non-concurrent or off line measures. A process or, more specifically, the processing of the L2 goes beyond paying mere attention to the linguistic features in the input (Gass, 1997; VanPatten, 2004).
When we report about learning as a product , we may be referring to what has been initially attended to or processed internally (e.g., intake) or the result or outcome of what has been further processed along the learning process (e.g., L2 knowledge) and demonstrated externally as output. At Stage 2, it is usually referred to as stored linguistic data kept in working memory at the stage of intake that is available for concurrent recognition and potential further process-ing but does not represent internalized or learned knowledge, which occurs further along the learning process. Typical assessment tests employed to measure intake include multiple-choice (MC) recognition or selection/identification tests. These tasks, usually administered off line, may differ in relation to the content of the items based on whether the content is true or not to the input provided or the number of options, which usually range from 4 (e.g., Leow, 2000) to 18 options (e.g., Godfroid et al., 2013). Tests of more than two options are usually considered more robust in relation to decreasing the potential of guessing.
Kind of Learning (Item Versus System Learning)
Intake may also be available for subsequent processing for potential incorporation into the developing internal system as systemized knowledge or un-systemized discrete items. If a learned product is hypothesized to reside at the internaliza-tion stage (Stage 4) of the learning process, that is, after the intake processing stage (Stage 3), then intake as a product at Stage 2 does not represent learning when viewed as L2 knowledge. This learned product is referred to at Stage 4 as stored knowledge that forms part of the developing grammatical system of the learner, potentially available for restructuring and subsequent language use. This stored knowledge may be accurate or inaccurate. If inaccurate, it is available to undergo further restructuring. Learner product or L2 knowl-edge in SLA is usually measured by off line assessment tasks after exposure to the L2 data. Tests designed to measure explicit knowledge include either oral (e.g., describe a series of drawings) or written (e.g., fill-in-the-blank) produc-tion tests, grammaticality judgment tests, off line verbal reports, and so on that require some visual or oral manifestation or grammatical description of the learned L2 knowledge. Tests employed to measure implicit knowledge include oral production, elicited imitation tests, and grammaticality judgment tests that are all timed to promote time pressure to encourage the use of feel rather than
rule and to reduce the opportunity to access metalinguistic knowledge (cf., e.g., Ellis, 2005). To measure un-systemized (item) and systemized knowledge, tests with old and new exemplars are employed, respectively. This learner product or L2 knowledge is typically what we test in our classrooms on quizzes and exams, especially if we ask our students to produce the L2 orally or in writing.
To address robust learning, it is always advisable to include delayed posttests to measure retention.
Type of Learning (Implicit vs. Explicit)
Learning can also be viewed as being implicit, that is, without awareness, or explicit, that is, with awareness. Type of learning, then, refers to whether the process of learning involved the construct of awareness. Awareness is defined by Tomlin and Villa (1994) as “a particular state of mind in which an individual has undergone a specific subjective experience of some cognitive content or external stimulus” (p. 193). Note that implicit or explicit learning is different from implicit or explicit knowledge , that is, the former takes place in Stages 1 and 3 while the latter resides in Stage 4 and is usually measured beyond Stage 5. Type of learning (implicit versus explicit) will be fully elaborated in Chapter 10 .
Measuring the Construct of Learning in SLA
Now with this tri-dimensional perspective of what comprises learning in this book in mind, which will form the foundation for my proposed model of the L2 learning process in Instructed SLA ( Chapter 12 ), the next step is to situate the experimental and assessment tasks employed in SLA studies to measure the con-struct of learning within the fine-grained framework of the L2 learning process presented in Chapter 2 .
Stages of the learning process in SLA: Of processes and products
INPUT
Regarding measurement of learning as a process or a product, once again the stage along the learning process from which data are elicited will indicate what type of process or product is being measured. With regard to learning as a process, be it item learning (exposure to old exemplars) or system learning (evi-denced by a task or test that contains new exemplars), there are currently three major concurrent data elicitation procedures being employed in the SLA field, namely eye-tracking, online verbal reports or think aloud protocols, and reaction time (these are discussed in more detail in Chapter 8) . At the input and intake processing stages, the three procedures may be used to gather data on learners’
processing and processes. As will be discussed later, whether processing or pro-cesses are the focus of the study will determine the selection of the concurrent data elicitation procedure.
With regard to learning as a product, there are several measures that can be employed to elicit data on learner attention or noticing reportedly paid (stimulated recalls), learner intake (e.g., offline MC recognition, interpretation), which would qualify as minimally addressing intake as a product held in working memory or lodged in the internal system as un-systemized discrete items, knowledge (e.g., GJT, fill-in-the-blank, etc.), and type of knowledge, whether implicit or explicit (e.g., offline verbal reports, confidence ratings, and source attribution).
Receptive Assessment Tasks
As can be seen, receptive tasks and tests are situated to address an early stage along the learning process, namely, the (processing of) intake held in working mem-ory (e.g., Leow, 2000) or in relation to some kind of knowledge already stored in the internal system (e.g., Leung & Williams, 2014). Receptive tasks and tests can be further divided into concurrent (e.g., interpretation/selection) and non-concurrent (e.g., offline MC recognition, interpretation). Let us take a closer look at two popular receptive tasks employed in current SLA research, namely, the MC recognition (e.g., Godfroid et al., 2013; Leow, 2000) and the MC interpretation (e.g., VanPatten & Cadierno, 1993; Williams, 2005) tasks.
Leow (2000) employed a four-option off line MC recognition task of old items immediately after exposure to the L2 data and reported that the aware group improved significantly from the pretest to the immediate posttest. Based on this assessment task alone, Leow could only report with some certainty that the ability to recognize the old targeted items was due to the presence of these items currently held in working memory (or episodic memory), and could also suggest or assume without hard evidence or much confidence that such items were also internalized as discrete items. However, a controlled written produc-tion test was also administered, and gain scores on this test were also significant.
These scores on a production test provided tangible evidence that targeted items were indeed internalized in the learners’ system, which allowed their produc-tion. However, neither assessment task (recognition or production) addressed
new exemplars, so this study only investigated item or un-systemized learning and could not report whether such learning was robust enough to remain over a longer period of time since no delayed posttest was administered. The inter-pretation of this study, then, is simple, as seen below:
new exemplars, so this study only investigated item or un-systemized learning and could not report whether such learning was robust enough to remain over a longer period of time since no delayed posttest was administered. The inter-pretation of this study, then, is simple, as seen below: