Incidental and Intentional Learning JAN H. HULSTIJN

1 Introduction There are two popular views on what it means to learn a second language. One view holds that it means months and even years of “intentional” study, involving the deliberate committing to memory of thousands of words (their meaning, sound, and spelling) and dozens of grammar rules. The other, complementary, view holds that much of the burden of intentional learning can be taken off the shoulders of the language learner by processes of “incidental” learning, involving the “picking up” of words and structures, simply by engaging in a variety of communicative activities, in particular reading and listening activities, during which the learner's attention is focused on the meaning rather than on the form of language. These popular views on intentional and incidental learning reflect, at best, only partially the ways in which these terms have been and are being used in the academic literature. Some empirical researchers attribute to them only a specific methodological meaning, in the context of laboratory-type learning experiments. Apart from this methodological sense, incidental and intentional learning have been given various interpretations, sometimes indistinguishable from two more widely used terms, namely implicit and explicit learning, respectively. There are virtually no experimental L2 grammar learning studies which are explicitly presented as “intentional” learning studies, and only a handful which are explicitly presented as studies on “incidental” learning. There is a vast literature, however, of empirical studies in incidental and intentional vocabulary learning. These empirical studies reflect a wide variety of theoretical and educational/pedagogic research questions; they, therefore, do not constitute a coherent research domain, as will become apparent in this chapter. The first aim of this chapter is to present the various ways in which the terms “incidental learning” and “intentional learning” are used in the psychological literature (section 2) and in the literature on L2 learning (section 3). The second aim is to give an overview of the empirical literature, in particular of the L2 vocabulary literature (section 4), as there are hardly any empirical studies on incidental and none on intentional L2 grammar learning (section 3.2). As the empirical literature on L2 vocabulary learning is so vast, and as the research questions differ so widely, section 4 confines itself to a number of illustrative examples. In section 5 follows a discussion of two pertinent methodological issues concerning the use of pre-tests and post-tests in incidental and intentional vocabulary learning studies. The chapter is concluded with some remarks concerning the diversity of issues addressed in it and the prospects of the labels “incidental learning” and “intentional learning” being used in the SLA field (section 6). Readers interested in the various meanings of incidental and intentional learning are advised to turn to section 3; readers interested in vocabulary learning may find section 4 most worthy of their attention, while methodologically oriented readers may be most interested in sections 2.2, 2.4, and 5. Boxes 12.1 and 12.2 give two examples of empirical research. The first study (Horst, Cobb, and Meara, 1998) illustrates how incidental vocabulary acquisition through reading can be investigated; the second study (Griffin and Harley, 1996) illustrates how an intentional design was used in a controlled study to investigate the role of various factors in learning a list of L2 words.1 These markedly contrasting studies are summarized in the boxes, and features not relevant in the present context have been omitted. 2. Incidental and Intentional Learning in the Psychological Literature In this section, the notions of incidental and intentional learning are traced back to their roots in psychology. First the rise of incidental and intentional learning is described in the era of stimulus- response psychology. This is followed by a methodological subsection, characterizing so-called Type I and Type II designs in experiments involving incidental and intentional learning. Then the fall and subsequent resurrection of incidental and intentional learning are described in the era of cognitive psychology. In the last subsection, the notion of transfer-appropriate processing, important for a proper understanding of learning experiments, is highlighted. 2.1 The origin of the notions of incidental and intentional learning in stimulus-response psychology According to early twentieth-century American psychologists such as James, Dewey, Watson, and Thorndike, learning is the forming of associations between sense impressions (stimuli – S) and impulses to action (responses – R). S-R psychologists distinguished various types of associative learning, ranging from elementary to complex (Gagné, 1965), but all involving the four basic concepts of stimulus, response, feedback, and conditioning. The most elementary form of learning is signal learning, requiring the making of a general, diffuse response to a stimulus (e.g., producing tears at the sight of onions). The next form in the learning hierarchy is stimulus-response learning (proper), requiring the making of a precise response to a discriminated stimulus. Learning L1-L2 word pairs is an example of stimulus-response learning. Sometimes, however, new words are learned through a series of S-R connections (so-called chains; more particularly, verbal chains, called verbal associations), as, for instance, when an English learner of French learns the L2 response allumette to the L1 stimulus match through the mediation of the English word illuminate and the word part lum, establishing the verbal chain match-illuminate-lum-allumette.2 According to psychologists at the time, an important determinant of the formation of associations (in human learning) is the apparent preparedness or state of readiness on the part of the learner, commonly referred to as set, intent, or motivation (Gibson, 1941; Postman and Senders, 1946; Underwood and Schulz, 1960). For many years, approximately from 1940 to 1965, psychologists tried to develop a theory of learning set, intent, or motivation.3 However, because of the difficulty of finding a satisfactory operationalization, researchers began to approach the concept merely in terms of the presence or absence of an explicit instruction to learn. The critical feature in this operationalization is whether or not (in incidental and intentional learning, respectively) participants are told in advance that they will be tested. Box 12.1 Incidental learning (Horst et al., 1998) Main research questions: i. Does reading a simplified novel lead to increased word knowledge? ii. Are words that occur more frequently in the text more likely to be learned? iii. Are words that occur more frequently in the language at large more likely to be learned? iv. Do learners with larger vocabulary sizes learn more words? Methodology: This was a one-group pre-test-treatment-post-test study of incidental L2 vocabulary learning. Subjects: 34 low-intermediate ESL learners in Oman (two intact classes), taking a reading course in preparation for the Cambridge Preliminary English Test. Task: The teachers read aloud a simplified version of Thomas Hardy's The Mayor of Casterbridge (109 pages; 21,232 words), while learners followed along in their books. This required six sessions, over a ten-day period. With the reading-aloud and reading-along procedure all subjects were exposed to the entire text, while creating “the circumstances for incidental acquisition by precluding opportunities for intentional learning” (p. 211). Students “appeared to be absorbed by the story of secret love, dissolution and remorse, and tears were shed for the mayor when he met his lonely death at the end” (p. 211). Students were pre-tested (about a week before the reading session commenced) and post-tested on their knowledge of 45 words of low and middle frequency levels, occurring between 2 and 17 times in the text. It was assumed that the one-week time lapse “would allow the items to be forgotten to the extent that they would not be immediately recognized as testing points when they were encountered in the story. This seems to have been effective; in a discussion held after the post-test, students were surprised to learn that the tested words had occurred repeatedly in The Mayor of Casterbridge. Their response also suggests that any word learning that occurred was implicit and incidental” (p. 213). Results: Mean vocabulary scores were 21.6 and 26.3 (out of 45) in pre- and post-test respectively (t [33] = 5.81; p < 0.05). Conclusions: Concerning the first research question, the authors conclude that these findings “offer conclusive evidence that small but substantial amounts of incidental vocabulary learning can occur as a result of reading a simplified novel” (p. 214), but also that “the power of incidental L2 vocabulary learning may have been overestimated” (p. 220). Concerning the three remaining research questions, sizable word gains are reported (i) when words occurred eight times in the text, (ii) when words (nouns) referred to concrete concepts, and (iii) when readers’ vocabulary size was at the (intermediate) 2000 level. Box 12.2 Intentional learning (Griffin and Harley, 1996) Research question: Is it more effective to learn word pairs in L1-L2 order or vice versa? This practical general question was broken down into the following sub-questions: i. Given a word pair A-B, is the association between the two components of the word pair bi- directional? ii. If it is bi-directional, is the forward association, A-B, stronger than the backward association, B-A? Is A more likely to lead to the recall of B than vice versa? (Use of forward association means being tested in the same direction as learning. Use of backward association means being tested in the opposite direction.) iii. Given that one component is familiar and the other is unfamiliar, is it more effective to learn the familiar-unfamiliar association (L1-L2) or the unfamiliar-familiar association (L2-L1)? iv. Is production or comprehension the easier task? (Production and comprehension in this context mean, respectively, giving an L2 item in response to an L1 item cue and giving an L1 item in response to an L2 item cue, irrespective of the direction of learning.) v. Does the direction of learning have an effect on remembering over time? One possibility considered was that, although the French-English bond might appear to be easier to establish, the English-French bond might be stronger over time, due to the initial difficulty of learning and its lack of list dependence. Methodology: Subjects: 47 and 63 students from two high schools in Britain, between 11 and 13 years of age, after six months of learning French. Task: Students were given 20 word pairs to learn, printed on a single sheet of paper. The instructions avoided the word “list” since the test would have the words in a different order from the original. Students were told that they would have eight minutes to learn the word pairs, that they would then hand back their papers and receive a written test. The test forms contained 20 words (either the English or French members of the learned word pairs); students had to write down the other member of each pair (cued recall). No instruction was given on either the learning technique or the mode of testing. In each school, four groups were formed. The arrangement of experimental groups is shown in table 12.1. Table 12.1 Arrangement of groups

The experiment adopted a 2 × 2 × 4 design, with two between-subject factors and one within- subjects factor. The between-subject factors were (i) use of forward or backward association at testing, and (ii) direction of learning (English-French or French-English). The within-subjects factor was time; students were tested four times: immediately after the learning session (day 1), as well as 3, 7, and 28 days later. No pre-test was administered. None of the French items had been encountered by students in their studies prior to the experiment, and students were not exposed to these words during the following 28 days. In order to answer the five research questions, performance of students in the following groups was compared: Research questions i and ii: comparison between groups 1 and 2 and groups 3 and 4. Research question iii: comparison between groups 1 and 3 and groups 2 and 4. Research question iv: comparison between groups 1 and 4 and groups 2 and 3. Research question v: a possible interaction between direction of learning and ability to recall over time. Results: The four groups of school B performed consistently lower than the groups of school A (grand means of 29 percent and 47 percent respectively of words correctly recalled). For simplicity's sake, only performance of school A groups will be reported here. For details, see the original study: Question i: The association was bi-directional: contra behaviorist claims, learning in one direction did not preclude performance in the opposite direction (37 percent in group 3 against 30 percent in group 4). Question ii: Forward association was stronger than backward association (60 percent mean scores in groups 1 and 2 against 34 percent in groups 3 and 4). Question iii: Direction of learning did not have a significant effect (45 percent mean scores for English-French learners in groups 1 and 3 and 48 percent for French-English learners in groups 2 and 4). Thus, there is nothing inherently more difficult about learning in the L1-L2 than in the L2-L1 direction. Question iv: Comprehension scores (52 percent in groups 2 and 3) were significantly higher than production scores (41 percent in groups 1 and 4). Question v: Performance on day 1 (53 percent) was significantly better than performance on day 3 (45 percent), day 7 (46 percent), and day 28 (43 percent). However, there was no significant interaction between language order at learning and day of testing. The English- French bond and the French-English bond decayed at much the same rate. Conclusion: The L1-L2 learning condition is, on balance, “the more versatile direction for learning when both production and comprehension are required” (p. 453). 2.2 Experimental operationalization of incidental and intentional learning: Type I and Type II designs In the heyday of S-R psychology, many studies were conducted to investigate the effect of a variety of manipulations of the stimulus materials, as well as of some learner variables such as age4. Two experimental methods were employed. The between-group Type I design is characteristic of the earlier studies. Participants in the incidental condition perform an orienting task on the stimulus materials, but they are given no instructions to learn and they are unexpectedly given a retention test afterwards. Participants in the intentional conditions are told in advance that they will later be tested. Early research aimed at demonstrating (i) that incidental learning did indeed exist and (ii) that intentional learning was superior to incidental learning. In the within-group Type II design, which was adopted in most later studies, all participants are instructed to learn some of the stimuli presented to them; but additional stimuli, which participants are not told to learn, are presented at the same time. Retention of the additional stimuli is unexpectedly tested afterwards. Thus, in the Type II design participants are their own controls, serving under both intentional and incidental conditions of learning, being exposed to two categories of stimuli, while expecting to be tested on only one of these. The additional stimuli in the Type II design may be either intrinsic or extrinsic, as illustrated with the following two hypothetical examples: Example 1, illustrating the use of intrinsic additional stimuli The stimulus materials contain target words, which are printed either in bold face or in italics and in either red or blue (yielding a 2 × 2 design of stimulus form). The orienting task focuses participants’ attention on color (instruction: “Try to remember which words appeared in red and which ones in blue”). Afterwards, participants are tested on their recall of red and blue words (intentional learning). But, unexpectedly, they are also requested to tell which words originally appeared in bold face and in italics (incidental learning). In this experimental design, the additional stimuli (typefaces) are said to be intrinsic because they belong to the same entities to which the attended stimuli (colors) belong. Example 2, illustrating the use of extrinsic additional stimuli The stimulus materials consist of a list of words some of which are printed in capitals and some in lower case. The orienting task focuses participants’ attention on the words in capitals (instruction: “Try to remember the capitalized words”). Afterwards, participants are tested on their recall of both capitalized (intentional) and lower-case words (incidental). The lower-case stimuli are said to be extrinsic to the experimenter-defined learning task, as they do not embody features of the attended stimuli. 2.3 Incidental and intentional learning in cognitive psychology With the decline of S-R psychology and the advent of cognitive psychology in the 1960s and 1970s, marking a fundamental paradigm shift, psychologists lost interest in the concept of set or intention as a central construct in the explanation of human learning and memory performance. This would have meant the demise of the constructs of incidental and intentional learning had not the work of some cognitive psychologists in the 1970s saved them from oblivion, not for theoretical but mainly for methodological reasons. Researchers of information processing and memory (the labels that replaced learning, which was felt to be associated too much with S-R psychology) in the 1970s, unearthed the Type II incidental learning design because it appeared to serve as an excellent tool in the investigation of the effect of various types of information processing on long-term information retention. For instance, in a seminal paper, Hyde and Jenkins (1973) presented groups of participants with a number of words and asked each group to perform a different orienting task. Participants were not told in advance that they would be later tested on their recall of the words. Jenkins and Hyde demonstrated that retention on the unexpected test fluctuated with orienting task. For instance, retention scores of participants who had rated the words as to their pleasantness or unpleasantness on a five-point scale (a semantic orienting task) were much higher than those of participants who had to record the part of speech of the words (a non-semantic orienting task).5 This and similar studies led Craik and Lockhart (1972) to propose their levels-of-processing theory, which engendered a lively theoretical debate and a great number of empirical investigations using incidental and (to a much lesser extent) intentional learning designs for many years to come (for a review, see Baddeley, 1997, ch. 7). It is through these studies that the notions of incidental and intentional learning have survived to the present day. For contemporary psychologists, their value is based on their record as research tools, rather than on their theoretical substance. In conclusion, incidental and intentional learning refer, strictly speaking, only to the absence or presence of an announcement to participants in a psychological experiment as to whether they will be tested after the experimental task. Thus, in the incidental case, the experiment may not even be explicitly presented as a “learning experiment,” because the word “learning” itself may already lead to testing expectancies among participants and hence to subject-generated information-processing strategies unwanted by the experimenter. In other words, incidental learning has acquired the status of a tool in the cognitive psychologist's experimental research kit to investigate some way or ways of information processing as intended by the investigator, not contaminated by ways of information processing not intended by the investigator. The presence or absence of an intention to learn does not figure as a theoretical construct in any current theory of human cognition. 2.4 Transfer-appropriate learning and the crucial role of the orienting task Retention or criterial tasks to be performed after a learning phase may be compatible, incompatible, or neutral to the processing mode of the previous learning task. In connection with this phenomenon of (in)compatibility between learning and retention task, Bransford, Franks, Morris, and Stein (1979) introduced the notion of transfer appropriateness. Bransford and his associates (Morris, Bransford, and Franks, 1977) found an interaction between encoding processes (semantic and non-semantic learning tasks) and the product of retrieval processes (semantic and non-semantic retention tasks). Participants who had been administered compatible learning and retention tasks (semantic-semantic, or non-semantic-non-semantic) achieved higher retention scores than participants who were given incompatible learning and retention tasks (semantic-non-semantic, or non-semantic-semantic). The lesson to be learned here is that an accurate assessment of intentional and incidental learning experiments requires a joint consideration of learning and retention task (Eysenck, 1982, p. 225).6 This can be illustrated with the study in box 12.2 (Griffin and Harley, 1996). In this intentional learning experiment, participants had to learn and memorize L2 words, which were paired to their L1 equivalents in either the L1-L2 productive order or in the L2-L1 receptive order. At test, the order was either the same as (productive-productive or receptive-receptive) or different from (productive- receptive or receptive-productive) the order during learning. It was found that retention scores on a same-order test were substantially higher than retention scores on a different-order test. The notion of transfer appropriateness may help to illustrate the difference between incidental and intentional learning. For example, as participants in an intentional vocabulary learning task are told in advance that they will be tested after the learning phase, they will try to store the word information that is to be learned in a form perceived as transferable to the test situation; and processing instructions during the learning phase in an incidental learning setting may or may not be conducive to successful transfer to the test situation. For instance, participants in an incidental learning vocabulary learning experiment who are instructed to pay attention to the meaning of some new words which appear in a reading text are likely to perform much better on an unexpected receptive post-test than on an unexpected productive post-test. The notion of transfer appropriateness also underscores the crucial importance of the orienting task given in a (Type I) learning study, because the orienting task is the instrument with which the researcher can control or manipulate participants’ attention to the information to be learned, and attention is a necessary condition for noticing and learning (Robinson, this volume; Schmidt, 2001).7 3. Incidental and Intentional Learning in the L2 Learning Literature This section will address the question of how incidental and intentional learning figure in the literature on L2 learning. As the field of L2 learning is fragmented into rather isolated sub-domains with little cross-talk, it comes as no surprise that the notions of incidental and intentional learning appear prominently in one domain but not at all in another. Incidental and intentional learning mainly figure in the area of vocabulary (including spelling). They do not appear at all in the areas of phonetics and phonology, however, and only exceptionally in the area of grammar (morphology and syntax).8 The reason why the term “intentional learning” does figure in the vocabulary learning literature but hardly in the literature on grammar learning, whereas “incidental” figures in both literatures, is that “incidental,” in principle, can apply to abstract as well as to factual declarative knowledge, whereas “intentional” appears to be applicable to factual knowledge only, as will be explained below in section 3.5.9 3.1 Weak theoretical interpretations of incidental learning Incidental learning has often been rather loosely interpreted in common terms, not firmly rooted in a particular theory. It could therefore be said to have several theoretical meanings, in the weak sense. From Schmidt (1994a) three definitions can be derived: i. The most general meaning is couched in negative terms as learning without the intent to learn (p. 16). ii. Another interpretation is that it refers to the learning of one stimulus aspect while paying attention to another stimulus aspect. As Schmidt (1994a, p. 16) puts it, incidental learning is “learning of one thing (...) when the learner's primary objective is to do something else (...).” This meaning of incidental clearly shows its descent from the methodological meaning, mentioned in section 2.1. iii. A slightly more specific interpretation of incidental learning is that it refers to the learning of formal features through a focus of attention on semantic features. Again, in the words of Schmidt (1994a, p. 16), but now with the previously omitted parenthetical phrases included: incidental learning is “learning of one thing (e.g., grammar) when the learner's primary objective is to do something else (e.g., communicate).” Recently, Gass (1999) suggested a new, extended meaning for incidental learning as the learning of grammatical structures without exposure to instances of these structures. She refers to two studies on the acquisition of relative clauses (Eckman, Bell, and Nelson, 1988; Gass, 1982) in which learners were exposed to some but not all types of relative clauses and appeared to have learned not only the structures presented to them but also, “incidentally,” the structures not presented in the input but implied by the ones that were presented. Thus, in addition to the distinction made in section 2.2 between intrinsic and extrinsic additional stimuli, and somewhat stretching the traditional notion of stimulus, one could even postulate a third category of implied, but not presented, and therefore not attended-to, stimuli. Most L2 learning researchers use incidental learning in connection with the learning of vocabulary through reading.10 As section 4.1 will show, it is widely believed that most people in literate societies enlarge their vocabularies through reading, focusing on the meaning of words and texts, rather than through the conscious, intentional memorization of lists of word forms and their meanings. A typical and well-known proponent of this view is Krashen (1989), who, in the context of his Input Hypothesis, argues that we acquire vocabulary and spelling through exposure to comprehensible input. 3.2 Empirical studies on incidental L2 grammar learning In many empirical L2 grammar-learning studies, participants are exposed to L2 data under various experimentally manipulated conditions, without being told that these data represent instances of some feature (principle or rule) of the L2 grammar and that the investigator's aim is to assess the extent to which participants are able to acquire this feature under the experimental conditions. It could be argued that, methodologically speaking, these studies are concerned with incidental learning. For example, in a well-known experiment, Doughty (1991) studied the acquisition of different kinds of English relative clauses by adult ESL learners. The study adopted a between- subjects design that included two experimental groups (and a control group, not relevant in the present context). Participants in one experimental group received meaning-oriented instruction; participants in the other experimental group received rule-oriented instruction. Neither experimental group was told in advance that they would be tested afterwards on their acquisition of various types of relative clauses. Thus, from a methodological perspective, both experimental groups can be called incidental groups. However, as the use of the term “incidental learning” would not have had a theoretically relevant meaning in this study, Doughty, understandably, found no reason to use this term.11 Only three experimental L2 grammar-learning studies appear to have explicitly used the term “incidental,” but none of them pitted incidental against intentional learning. The first study (Hulstijn, 1989) involved three experimental groups (Form, Meaning, and Form and Meaning). Theoretically, the study is presented as one of implicit learning, meaning that learners were not consciously aware of the grammatical target features under investigation. Methodologically, the study is presented as an incidental learning study: “Ss were not informed about the research questions until after the completion of the last test, and, while carrying out a current task, did not know whether a subsequent test would follow” (p. 54). The second and third studies (reported, respectively, in Robinson 1996a, 1996b, 1997) involved four experimental conditions: Implicit, Incidental, Explicit Rule Search, and Explicit Instruction (in the 1996 study), and Implicit, Incidental, Enhanced, and Instructed (in the 1997 study). The implicit and incidental conditions were alike “in not requiring a conscious focus on the grammatical form of the stimuli presented during training. In the implicit condition the task instruction is to memorize sentences, whereas in the incidental condition the task instruction is to read the sentences and understand their meaning” (Robinson, 1996b, p. 35). Robinson, who motivates his use of the term “incidental” by a reference to Paradis (1994, p. 394), whose definition will be quoted in section 3.4, appears to use “incidental” to refer to learning of L2 forms through a focus of attention on meaning, as in the third definition mentioned in section 3.1. In summary, although some definitions of incidental L2 grammar learning have been proposed in the literature (in particular, Gass, 1999; Schmidt, 1994a), no reports of empirical L2 grammar learning studies have so far been published which explicitly base themselves on the Schmidt or Gass definitions. This is quite understandable, as the notion of implicit learning has had a greater appeal among SLA researchers than the notion of incidental learning (see section 3.4; DeKeyser, this volume; Doughty, this volume).12 3.3 The meaning of “intentional” in the vocabulary-learning literature In the literature on vocabulary learning, when used at all, “intentional learning” is commonly given a cognitive interpretation, as the rehearsal and memorizing techniques invoked by learners when they have the explicit intention of learning and retaining lexical information (Schmitt, 1997). 3.4 The differences between incidental and implicit and between intentional and explicit types of learning For many authors, incidental and intentional learning overlap with, or even become indistinguishable from, implicit and explicit learning respectively. There are several interpretations of the terms “implicit” and “explicit” learning (see DeKeyser, this volume). The most characteristic feature, however, distinguishing implicit from explicit learning is the absence or presence of “conscious operations” (N. Ellis, 1994, p. 1), a notion also referred to as the absence or presence of “awareness at the point of learning” (Schmidt, 1994a, p. 20). Note that none of the definitions of incidental and intentional listed in section 3.1 is synonymous with the definitions of implicit and explicit learning given by Ellis and Schmidt. In line with Schmidt (1994a), it is recommended here that the distinctions between incidental and implicit and between intentional and explicit should be maintained. Paradis (1994, p. 394), for instance, distinguishes incidental from implicit in the following definition of implicit competence, which “is acquired incidentally (i.e., by not focusing attention on what is being internalized, as in acquiring the form while focusing on the meaning), stored implicitly (i.e., not available to conscious awareness), and used automatically (i.e., without conscious control).” Thus, incidental learning, in all the definitions listed in section 3.1, is always implicated in implicit learning; implicit learning thus entails more than what is meant by incidental learning. In a similar vein, it is recommended here that a distinction be maintained between intentional and explicit learning. Whereas explicit learning involves awareness at the point of learning (e.g., by trying to understand what the function of a certain language form is), intentional learning involves a deliberate attempt to commit new information to memory (e.g., by applying rehearsal and/or mnemonic techniques). 3.5 Confusions concerning the interaction of the what and how of incidental and intentional learning The nagging problem in discussions concerning incidental and intentional (as well as implicit and explicit) learning is that, although the definitions of these terms appear to refer to the how of learning (learning mechanisms), their interpretations depend on authors’ views on the what of learning (the representation of knowledge in the mind/brain).13 For instance, it is relatively easy to imagine the intentional learning of a list of L2 words, as these form-meaning connections are readily conceived of as instances of declarative, factual knowledge. However, as soon as we define the what of learning as abstract knowledge of properties of L2 grammar (e.g., knowledge of the L2 setting of the pro-drop parameter), it is almost impossible to conceive of the acquisition of this abstract grammatical feature taking place through intentional learning. It is much easier, it seems, to conceive of the acquisition of this feature taking place through implicit, and hence through incidental learning (see section 3.4). This and similar interactions between the what and how of L2 learning have caused, and continue to cause, confusions in the L2 learning literature.14 It comes as no surprise, therefore, that the area in which “incidental” and “intentional” are used most frequently is that of vocabulary learning. Vocabulary knowledge can easily be conceived of as a type of declarative knowledge, and it is declarative knowledge which can be learned intentionally (e.g., with various memory aids) as well as incidentally (e.g., through reading and listening). It can be concluded that incidental and intentional learning are differentially important for different classes of target language features: whereas incidental is used in connection with the learning of both abstract and factual knowledge, the use of intentional is restricted to the learning of factual knowledge. When used in connection with factual knowledge, incidental and intentional learning in the realm of language (e.g., learning vocabulary items, writing systems, spelling rules, conventions for addressing people in oral or written discourse according to their age, sex, and status) does not appear to differ from incidental and intentional learning in other walks of life (e.g., learning geographical names, historical events). 3.6 The issue of two poles on a continuum as opposed to two distinct categories Should incidental and intentional learning be thought of as two distinct learning processes or as poles on a continuum? There is no simple answer to this question. As Schmidt (1994a, 1994b) has argued, there is no learning without attention and noticing. This is true not only for implicit but also for incidental learning. Incidental and intentional share the involvement of attention and noticing (see the quotation from Paradis, 1994, p. 394, in section 3.4). Thus, in the dimension of attention and noticing, incidental and intentional do not form two distinct categories. However, this still leaves open possibilities of distinct processes in other dimensions. As was mentioned and illustrated in section 3.2, incidental and intentional are not juxtaposed to each other in the L2 grammar-learning literature. The polarity issue, therefore, does not seem to play a role in the domain of grammar learning. In the L2 vocabulary-learning literature, however, incidental and intentional learning are seen as distinct categories, in that intentional learning does, and incidental does not, imply the use of deliberate retention techniques. In conclusion, on the one hand, both incidental and intentional learning require some attention and noticing. On the other hand, however, attention is deliberately directed to committing new information to memory in the case of intentional learning, whereas the involvement of attention is not deliberately geared toward an articulated learning goal in the case of incidental learning.