• No results found

CHAPTER 2: REVIEW OF THE LITERATURE

2.4 Individual Differences in Cognitive Ability

2.4.6 Aptitude as a Variable Interacting with Learning Conditions

Robinson’s (2001a, 2002, 2007a, 2012) Aptitude Complexes Hypothesis provides underpinning for research on aptitude-treatment interaction. Robinson’s model comprises four aptitude complexes that involve cognitive processes related to primary abilities such as perceptual speed, pattern recognition, phonological working memory capacity, speed of processing in phonological working memory, working memory for text and speed of working memory for text, analogies of meaning and inferring word meaning,

135

grammatical sensitivity and rote memory. Robinson argues that these primary abilities entail second-order abilities, such as noticing the gap, memory for contingent speech, memory for contingent text, deep semantic processing and metalinguistic rule rehearsal (see Robinson, 2001a, 2002, 2007a, 2012). Robinson suggests that these abilities, or a combination of them, promote L2 development but their facilitative role is linked to specific learning conditions. For example, in Robinson’s model, the extent to which learners notice and benefit from recasts is associated with (1) ability to notice the gap between an erroneous utterance and its reformulation and (2) memory for contingent speech (i.e. a reformulation has to be held in memory while comparing it with erroneous output). Robinson also expects that incidental learning from oral content involving a flood of linguistic features is related to memory for contingent speech and deep semantic processing, while incidental learning from written content is linked to memory for contingent text and deep semantic processing. Finally, explicit rule learning combined with written examples could be influenced by memory for contingent text and metalinguistic rule rehearsal.

In the second section of his framework, i.e. the Ability Differentiation Hypothesis, Robinson suggests that there is variation among L2 learners regarding their strengths and weaknesses in the cognitive ability involved in L2 aptitude. Due to these variations, learning conditions should be adapted to learners’ ability in order to arrive at superior L2 outcomes. For example, recasting might be an effective intervention for learners with strengths in noticing the gap and memory for contingent speech, while for learners

136

lacking these strengths, alternative focus-on-form techniques may be beneficial. Several empirical studies have sought to elucidate whether there is a relationship between individual differences in L2 aptitude and the effectiveness of different types of instruction (DeGraaff, 1997; Erlam, 2005; Hwu & Sun, 2012; Robinson, 1997; VanPatten, Collopy, Price, Borst, & Qualin, 2013).

Robinson (1997) explored the extent to which the potential benefits of four learning conditions are influenced by L2 aptitude measured by the MLAT (Carroll & Sapon, 1959). The learning conditions were classified into instructed condition, rule-search condition, implicit condition and incidental condition. In the instructed condition, first, the learners were provided with an explanation of rules, and second they were required to apply these rules to sentences (e.g. to answer questions regarding metalinguistic information about form). In the rule-search condition, the participants were instructed to look at sentences and find rules by themselves. In the implicit condition, the learners were presented with sentences and asked questions by the researcher (e.g. about the location of words), but without being given with any explicit or metalinguistic information. Finally, in the incidental condition, the learners read sentences and had to answer comprehension questions related to their content. The participants received feedback about the correctness of their answers in all of the conditions, apart from the rule- search condition where their answers could not be predicted. L2 learning was measured by a GJT and rule awareness was examined by a questionnaire asking the participants whether they had noticed or searched for rules and

137

could verbalize them. Robinson found that all the experimental groups were influenced by aptitude, apart from the incidental learning group.

Similar to Robinson (1997), DeGraff (1997) also employed different instructional conditions in order to investigate their potential relationship with aptitude. Aptitude was operationalized as grammatical sensitivity and rote memory and it was measured by a Dutch version of the MLAT test and learners’ ability to infer the meanings of novel words from the context. The L2 was an artificial language called eXperanto. In the explicit group, the learners’ attention to linguistic elements was drawn by providing them with grammatical explanations and highlighting the target constructions. In contrast, no metalinguistic explanation was offered under the implicit condition. L2 learning was assessed by GJTs with and without time pressure and an untimed gap-filling task. DeGraff found significant correlations between L2 aptitude and both groups’ performance on an immediate and a delayed posttest.

Erlam (2005) examined the relationship between individual differences in language analytic ability, phonetic coding ability and working memory and three different types of instruction, namely, deductive instruction, inductive instruction and structured input instruction addressing direct object pronouns. Deductive instruction was coded as an explicit learning condition because it entailed rule explanation and learners’ engagement with form- focused activities requiring the production of the target feature without time pressure. Hence, the participants had time to use the rules explained

138

to them prior to the activities. These activities were followed by CF that aimed to draw their attention to the rules of the target construction. Inductive instruction was coded as an implicit learning condition. The participants performed activities and were expected to make their own hypotheses about the target feature without being presented with rules or any metalinguistic information. During structured input instruction, the learners were also presented with rules about the target feature, followed by input-based activities and consciousness-raising activities. The former required the participants to process both spoken and written input, including the target construction, whereas the latter asked them to identify errors. Neither of these activities involved production of the target feature. Erlam found that the explicit deductive instruction was less affected by aptitude, as opposed to (1) inductive instruction and (2) structured input instruction, both of whose benefits were moderated by the participants’ L2 aptitude.

In a similar vein, Hwu and Sun (2012) examined the relationship between L2 aptitude and the effectiveness of two types of explicit instruction in facilitating the development of the Spanish verb gustar. The aptitude constructs they explored were analytical language ability and associative memory, measured by the MLAT and memory for text. The researchers employed two instructional conditions involving deduction and induction coded as explicit. Both groups were provided with an explanation of rules and metalinguistic information through instructional activities in the deductive condition, and through multiple-choice questions and

139

metalinguistic feedback in the explicit inductive condition. L2 development was assessed by a pretest, an immediate posttest and a delayed posttest composed of tasks that required written sentence production and written sentence correction. The study indicated that the L2 gains of both groups were influenced by aptitude, and particularly by their memory for text. Finally, VanPatten et al. (2013) conducted three experiments that differed with regard to the L2 in focus (i.e. Spanish, German and Russian for each experiment, respectively). The aim was to illuminate the relationship between the participants’ grammatical sensitivity measured by the MLAT, the learners’ processing time (i.e. length of time they needed in order to start processing the target feature accurately) and their improvement on the target construction. The linguistic area addressed referred to the flexibility of the target languages with respect to the order of words functioning as subjects or objects and whether they afforded both subject-verb-object (SVO) constructions and object-verb-subject (OVS) constructions. Based on the First-Noun Principle, the learners were expected to process the first noun as subject in both types of sentences. That would be the target-like option for the SVO constructions but not for the OVS ones. The participants were assigned into two groups, both receiving structured input of the target linguistic element; however, the two experimental conditions differed as to whether the structured input was combined with explicit information or not. The researchers found that the degree to which the students benefited from structured input positively correlated with their grammatical sensitivity only under the condition that involved explicit information. It should also be

140

noted that this finding was demonstrated only in one of the three experiments, the one utilizing German as L2.

Overall, the studies presented above have shown that instructional conditions, regardless of whether they are coded by the researchers as explicit or implicit, are related to L2 aptitude. That is, learners with greater aptitude are likely to benefit more from both explicit and implicit instruction. However, these findings should be interpreted with caution. The factor that differentiates explicit from implicit learning is awareness; learning is implicit when learners are not aware of what they have learnt and explicit when awareness is implicated (DeKeyser, 2003; Rebuschat, 2013). Hence, an instructional condition devoid of the provision of rules and metalinguistic information does not necessarily entail implicit learning, as although the learners might not be able to verbalize a rule, they might have employed explicit cognitive mechanisms during the learning process and could be aware of their emerging L2 knowledge. Consequently, the instructional conditions operationalized as implicit in the studies above cannot exclude the possibility of explicit learning being involved, and they cannot provide robust evidence about the relationship of L2 aptitude with purely implicit vs explicit learning.

Another interesting parameter that should be taken into account is the impact of aptitude on the acquisition of linguistic constructions that differ in terms of salience, redundancy and/or rule complexity (Skehan, 2014a; Yalçın & Spada, 2016). For example, Yalçın and Spada (2016) examined the extent to which aptitude is associated with the acquisition of two linguistic areas

141

that differed in rule complexity. The participants received four hours of instruction on the passive and the past progressive, with the former being coded as a more difficult structure than the latter. L2 gains were assessed by a pre-test and a post-test composed of written tasks, untimed GJTs and oral production tasks. L2 aptitude was measured by the LLAMA Aptitude test (Meara, 2005). Interestingly, the study revealed that different components of aptitude contributed to development of the two constructions. In particular, greater grammatical inferencing facilitated benefits related to the passive, whereas greater associative memory assisted improvement in the past progressive.

Within the framework of aptitude-treatment interaction, researchers have also explored the relationship of aptitude with L2 gains resulting from different types of CF (Li, 2013, 2015; Sheen, 2007; Trofimovich, Ammar, & Gatbonton, 2007; Yilmaz, 2013; Yilmaz & Koylu, 2016; Yilmaz, Granena, & Meyer, 2016; Yilmaz & Granena, 2015). These studies are discussed in greater detail in the following section.