Effects of differences in language aptitude on learning grammatical collocations under elaborated input conditions

Abstract

The purpose of this study was to find out, whether three methods of presenting input, were effective in relation to language aptitude. Persian-speaking learners of English were provided with 20 grammatical collocations (verb–preposition collocations) embedded in authentic passages, lexically/grammatically elaborated passages, and lexically/grammatically elaborated passages with shorter sentences and enhanced target collocations. Participants were assessed on their receptive knowledge and productive knowledge of the grammatical collocations by a posttest and a delayed posttest, and their scores were correlated with various measures of language aptitude. The results suggested that modified elaborated input, which provided more focus on form (FonF), positively affected immediate and long-term gains in receptive knowledge and ruled out individual differences (IDs) in language aptitude. The results also implicated working memory (WM) as an explanatory variable in immediate and long-term achievements in productive knowledge under all the conditions of presenting input.

Keywords

elaborated input genuine input grammatical collocation language aptitude modified elaborated input phrasal verbs

I Introduction

An important concern in the field of second language acquisition (SLA) has been the necessity of uncovering individual differences (IDs) in the process of learning. Second language (L2) learners do not benefit from different instructional methods uniformly; thus, SLA research has become concerned with the relative effectiveness of L2 instructions when they are matched with IDs (Yalçın & Spada, 2016). Regarding differences in L2 learners’ cognitive abilities, Aptitude–Treatment–Interaction (ATI) studies deal with the role that language aptitude plays in mediating the effects of instructional treatments on the cognitive processes that facilitate or impede language acquisition (R. Ellis, 2012). As a research paradigm, ATI aims to find out the crucial role of aptitude for language learning, especially in the post-critical period (Granena & Long, 2013). Some of the studies in the field of ATI are fully crossed in the sense that distinct groups are set up based on L2 learners’ aptitude profiles and then are matched and mismatched with particular treatments. In other ATI studies, aptitude measures are treated as continuous variables (in contrast with grouping variables) to gauge the relationship between L2 learners’ language aptitude and the gains resulting from instructional treatments (Vatz et al., 2013). In the present study, which used intact classes, the second design was employed.

II Language aptitude

Language aptitude has been defined by Granena (2013) as a ‘combination of cognitive and perceptual abilities that are advantageous in SLA’ (p. 665). In his first model of aptitude, Carroll (1981) proposed that four, relatively independent, aptitude components represent abilities for foreign language learning: (1) phonetic coding ability (PCA) (the ability to recognize different sounds, associate them with their representative symbols, and remember these associations); (2) grammatical sensitivity (the ability to recognize the grammatical functions of words in sentences); (3) inductive language learning ability (the ability to infer language rules from a given corpus); and (4) rote learning ability, or associative memory (the ability to learn the associations of sounds and meanings and remember these associations). Carroll developed the Modern Language Aptitude Test (MLAT) to measure these components (Carroll & Sapon, 1959), with the exception of inductive language learning ability. Carroll’s conceptual framework of foreign language aptitude has survived in the long run, even better than the MLAT, and it is still a robust representation of foreign language aptitude (Skehan, 2012).

The MLAT was developed prior to more recent SLA research on aptitude demands of implicit learning (Robinson, 2012) and working memory capacity (WMC) (e.g. Conway et al., 2005). As an impermanent cognitive workspace, working memory (WM) includes storage and processing of language (Juffs & Harrington, 2011). A well-known WM model by Baddeley (2000) is comprised of a multi-component system, where a ‘central executive’ organizes the information that flows between three specialized domains: the phonological loop, the visuospatial sketchpad, and the episodic buffer. The phonological loop deals with phonological information, the visuospatial sketchpad handles visual information, and the episodic buffer is a place for integrating and reusing the information. The phonological loop is generally indexed by non-word recognition (Juffs & Harrington, 2011). If non-words are based on the syllable structure of the L2, not first language (L1), phonological non-word tests can adequately represent PCA (Skehan, 2012). Measures of WMC evaluate the function of the central executive and are different and independent from measures of phonetic capacity (Wen, 2015).

Integration of WM makes language aptitude research accord with contemporary cognitive psychology. Research studies suggest that implicit processing can necessitate employing WMC by L2 learners. Sáfár and Kormos (2008) investigated the relation of language aptitude with a communicative language teaching with focus on form (FonF) instruction and found WM to be a key cognitive variable affecting learning. In another study, Indrarathne and Kormos (2017) showed that WM was closely related to receptive knowledge of a grammatical construction in implicit learning contexts with input flood and input enhancement. However, productive knowledge was not strongly influenced by WMC in the implicit conditions.

Following developments in psycholinguistics and cognitive science, Robinson (2012) and Skehan (2002, 2012) have further emphasized the role of language aptitude as a combination of cognitive abilities. They proposed that different L2 learners may have different aptitude profiles when the occasion arises; some learners in a group, for example, may have high WMC while other members of that group may be better at analyzing language. Robinson and Skehan developed two models of language aptitude that may help to tailor L2 learners’ traits and needs to L2 instruction (Yalçın & Spada, 2016). In their models, they underlined the important role of WM, especially in FonF conditions in which L2 learners’ attention is drawn to features (grammar or lexis) while they are attending to meaning. In Robinson’s Aptitude Complex/Ability Differentiation framework, language aptitude is a multi-componential construct, and it is implicated in various L2 learning contexts. Robinson proposed that language aptitude is comprised of primary abilities which combine into high-order ability factors. These ability factors, in turn, make aptitude complexes, or combinations of cognitive abilities, differentially involved in acquisitional processes of different FonF conditions of exposure to L2 input. Among these L2 learning conditions, incidental learning via written input was associated with the aptitude complex of WM and deep semantic processing.

While in Robinson’s model conditions of L2 instruction were the operational platform, Skehan’s (1998) Processing Stage framework carried possible implications for matching language aptitude components to input, central processing, and output stages of SLA acquisition and their corresponding cognitive processes. Skehan (2002, 2012) identified various L2 cognitive processes involved in learning: input processing, noticing, pattern identification, extension, complexification and restructuring, integration, error avoidance, automatization, repertoire and silence creation, and lexicalization. In his Processing Stage model, the first cognitive process concerns the stage of input processing, the next five processes emphasize central processing and interlanguage development stages, and the final four processes, from error avoidance to lexicalization, deal with using language in output. Skehan argued that input processing and noticing pertain to PCA and WM. Moreover, perceiving patterns in input and restructuring them (for further generalization), and extending these patterns (to suit meaningful contexts) are linked to language analytic ability (LAA) and WM.

III Aptitude–treatment interaction

As already mentioned, different types of language instruction encourage different types of processing that rely on aptitude components. In a study that compared the effects of deductive instruction, inductive instruction, and structured input instruction on the acquisition of direct object pronouns, Erlam (2005) applied Skehan’s (1998) processing framework. She selected PCA, LAA, and memory to be measured with the Sound Discrimination test of the Pimsleur Language Aptitude Battery (PLAB) (Pimsleur, 1966), the Words in Sentences subtest of the MLAT, and a test of memory, respectively. The aptitude measures were correlated with the learners’ listening comprehension, reading comprehension, written production, and oral production outcomes. The results indicated that IDs were neutralized with the effect of deductive instruction. In another study, Nagata, Aline, and R. Ellis (1999) examined the impact of language aptitude on comprehension of modified input (oral directions in a non-reciprocal listening task) and incidental acquisition of infrequent words incorporated into modified input. They correlated the results of five aptitude tests with comprehension and word-acquisition scores. The aptitude tests, associated with the component model proposed by Skehan, included the Sound Discrimination and Sound–Symbol Association subtests of the PLAB, Words in Sentences subtest of the MLAT, and a Finnish memory test along with the Paired Associate subtest of the MLAT. The results indicated that the components of language aptitude contributed to comprehending oral directions, and only memory ability to the acquisition of new lexical items. Yalçın and Spada (2016) supported Skehan’s (2002) proposal that different stages of language acquisition may relate to different components of aptitude. Using a quasi-experimental design, they explored the relationship between learning difficult and easy English structures and the components of the LLAMA (Meara, 2005), which is an aptitude test modeled on the MLAT. The results demonstrated that L2 learners relied on their memory ability to deal with easier structures.

Many studies have focused on the acquisition of language rules by investigating the dynamic interplay between language aptitude and explicit or implicit learning conditions (e.g. Granena, 2013; Li, 2013; Robinson, 2002; Sheen, 2007; Yilmaz & Granena, 2016). The few studies exploring the acquisition of L2 collocations have concerned the interaction of very advanced L2 learners’ aptitude components with learning in mostly natural contexts. For example, Forsberg Lundell and Sandgren (2013) investigated the relationship between language aptitude, as measured by the LLAMA, and production of L2 French verb–noun collocations. The results indicated that the scores were positively related to the aptitude measures, including grammatical inferencing and sound–symbol correspondence, but only the correlation between the test scores and the sound recognition subtest of the LLAMA was significant. Forsberg Lundell and Sandgren’s findings were compatible with Granena and Long’s (2013) results that revealed a significant correlation between scores on the auditory subtests of the LLAMA and the acquisition of L2 collocations. Even though these studies expand our knowledge, questions still remain about the role that language aptitude might play when L2 learners encounter collocations in meaningful learning contexts in L2 classrooms.

IV Background to the study and research questions

Long (2015b) pointed out that one lacuna in ATI concerns the performance of L2 learners with different cognitive abilities when they encounter grammatical collocations in elaborated input. This study follows Long’s suggestion by examining how various aptitude components are associated with the learning of collocations through three types of L2 input that not only differ in their degree of linguistic complexity but also their level of implicitness. In this study, implicit learning was operationalized as selective attention to some aspects of the input for promoting FonF when L2 learners are learning incidentally (Granena, 2013).

Another aim of the present study is to shed light on the interplay between the aptitude components in the Processing Stage model and learning through L2 written input. SLA research has been primarily concerned with associating phonetic capacity with analyzing aural input (Forsberg Lundell & Sandgren, 2013). There is growing evidence that phonological processing skills are central to approaching L2 written input as well, because for word recognition to happen, readers need to access lexical entries in their mental lexicon that include phonological, orthographic, semantic, and syntactic information (Grabe, 2009). As a result, readers who code letter-sound links more efficiently may be better able to activate phonological forms for successful word recognition and thus comprehension during reading.

Probing the interaction of implicit-FonF input conditions with language aptitude can be both theoretically and practically important. Theoretically, it might be necessary to know the construct validity of different aptitude components hypothesized to have major roles at different stages of acquisition. At the practical, pedagogical, decision-making level, it is important to match learners’ aptitude profiles to appropriate treatments where possible. In the present study, the treatments contained authentic passages (genuine input), lexically/grammatically elaborated passages (elaborated input), and lexically/grammatically elaborated passages with typographically enhanced collocations and optimized sentence length (modified elaborated input). These types of input were deemed to attract L2 learners’ attention to the forms and meanings of English grammatical collocations to different degrees. Based on the preceding theoretical considerations, the study was designed to address the relationship between three learning conditions (namely, genuine input, elaborated input, and modified elaborated input) and Skehan’s (2002, 2012) main components of aptitude (i.e. PCA, LAA, and WM):

To what extent is language aptitude related to L2 learners’ acquisition of L2 collocations when they receive genuine input?

To what extent is language aptitude related to L2 learners’ acquisition of L2 collocations when they receive elaborated input?

To what extent is language aptitude related to L2 learners’ acquisition of L2 collocations when they receive modified elaborated input?

V Context and methodology

1 Participants

The present study was conducted in a language learning institute for Iranian high school students aged between 14 and 18 years old. They received four hours of English instruction per week, which mainly focused on communicative skills. Additionally, for all of them, three hours of learning English was obligatory at high school. Eighty-seven L2 learners of English participated in this study. Data from four participants who were absent during the treatment sessions or the posttests were removed. Participants (n = 83) had received two years of English instruction at the language institute and were considered to be at an intermediate level of English proficiency according to the Oxford Placement Test and their teacher’s ratings. Participants gave their consent to participate in the study during their English course.

2 Design

This quasi-experimental pre-/post-/delayed posttest intervention study took place with four intact classes, which were selected as the four study groups: genuine input (GI) group (n = 22), elaborated input (EI) group (n = 22), modified elaborated (MEI) group (n = 21), and a control group (n = 18). Measures of receptive knowledge and productive knowledge of the target items were given as a pretest, posttest, and delayed posttest to all the groups. These tests were administered in the same versions. The purpose of the pretest was to determine whether participants knew the target grammatical collocations, and it was administered two weeks before the treatment in order not to attract L2 learners’ attention to the target items. The treatment spanned three weeks (within five treatment sessions), and was followed by the posttest (five days after the treatment) for finding the immediate gains, and the delayed posttest (three weeks after the treatment) for finding the long-term gains. As explained later, both the posttest and delayed posttest contained a productive knowledge and a receptive knowledge test. Based on the instructor’s experience with a group of intermediate L2 learners who had been administered receptive and productive collocation knowledge tests in a language institute, participants in the present study were given 20- minute and 15-minute time limits to complete the productive knowledge test and the receptive knowledge test, respectively. Participants in the treatment groups were administered the posttest and delayed posttest at the same time and place. During the testing sessions, most of the participants (more than two-thirds) completed the collocation knowledge tests a few minutes before the test time ended. To insure higher reliability, other participants also were prompted by their instructor to hand in their test papers because when they were observed, they were not using their time for performing the test items. Polio, Fleck, and Leder (1998) state that extra test time that does not involve participants in linguistic processing may decrease test reliability.

For the treatment groups (n = 65), the aptitude measures provided the independent variables, and the scores on receptive and productive knowledge posttests and delayed posttests presented the dependent variables. The control group followed the normal curriculum of the language institute without receiving any of the specific types of input but took the pretest, posttest, and delayed posttest at the same time as the other groups (for a schematic representation of the study procedure for the treatment and control groups, see Figure 1).

Figure 1.

Schematic representation of the study procedure.

The GI, EI, and MEI groups received genuine input, elaborated input, and modified elaborated input, respectively, by their regular instructor in their usual 120-minute class sessions. The target items to learn were 20 grammatical collocations (see Appendix 1). Each session, L2 learners in the treatment groups encountered four collocations. Each collocation was embedded in five passages that were presented together. The passages had been produced in genuine, elaborated, and modified elaborated versions, as explained later. In order to encourage meaningful interaction with the texts, each passage was followed by a statement, which included the target collocation in the passage, and tapped L2 learners’ general comprehension. Indeed, L2 learners were required to determine that the statement was true or false (for an example, see Appendix 2).

The time to read the passages did not vary between the groups in order to determine the effect of the treatments rather than time-on-task differences (Long, 2015a). By piloting the three versions of some of the passages with L2 learners similar to the main participants, an average of two minutes was specified for reading each passage. Therefore, the time spent on each collocation (embedded in five passages) was 2 × 5 minutes, and on four collocations thus 4 × 2 × 5 minutes. In order to prevent L2 learners’ fatigue, after each 10 minutes dedicated to the treatment, the class conventional activities were performed. Table 1 shows the time and the activities in instructional sessions.

Table 1.

Time management in each treatment session.

Time (120 minutes)	Activity
10 minutes	First treatment
20 minutes	Class routine
10 minutes	Second treatment
20 minutes	Class routine
10 minutes	Third treatment
20 minutes	Class routine
10 minutes	Fourth treatment
20 minutes	Class routine

Since participants had to read five passages (related to one target collocation) and their comprehension statements in 10 minutes, they were instructed to read each passage and its following statement within two minutes. They were not told about the collocation knowledge posttest and delayed posttest they were going to take.

3 Target items

Twenty infrequent grammatical collocations, with around 60 occurrences in the Corpus of Contemporary American English (COCA), were selected as the target items. We selected infrequent collocations in order to reduce the likelihood that these collocations were already known, or at least being experienced, by the participants before the study. Distinguished from lexical collocations, grammatical collocations are the co-occurrence of lexical words and grammatical words; they are the combination of a dominant word such as verb, noun, or adjective and a preposition or grammatical structure (Barfield, 2013). In the present study, the target grammatical collocations were verb–preposition combinations, also commonly referred to as phrasal verbs. Long (2007) also suggested using this type of collocations as target stimuli in genuine and elaborated reading materials. The frequency of verb–preposition co-occurrences, without possible noun phrases or pronouns between them, was the criterion for their selection.

4 Materials

Both receptive and productive knowledge tests were used. The content validity of the tests was endorsed by three experts with a specialization in L2 teaching. As measures of the main components of Shehan’s model of language aptitude, L2 learners completed four aptitude tests. Although some SLA researchers (e.g. Robinson, 2012) have expressed skepticism over the measures of LAA as predictors of learning from purely incidental conditions, employing these measures as potential predictors of success under implicit learning conditions has been clearly justified (Li, 2013; Robinson, 2002; Skehan, 2002).

a Phonetic coding ability test

The PLAB emphasizes auditory factors by including the Sound Discrimination and the Sound–Symbol Association subtests that assess PCA (Nagata et al., 1999). For taking the Sound Discrimination test, some words should be learned in a rare language and have to be told apart in different contexts. The Sound–Symbol Association test measures the ability to link graphological and phonological forms of tape-recorded, nonsense words such as thurskle. Participants were tested with the Sound–Symbol Association test as a way of evaluating PCA in this research. Using Cronbach’s alpha, reliability of the test for the participants (n = 65) was α = 0.72 (M = 18.10, SD = 14.20).

b Language analytic ability test

The Language Analysis subtest of the PLAB is similar to the Words in Sentences subtest of the MLAT, but it measures inductive language ability. In the present study, the Language Analysis subtest of the PLAB was used, first, because the initial L2 instruction that all the participants of the present study had received at their schools was based on grammar. According to Harley and Hart (1997), who preferred the Language Analysis subtest of the PLAB over the Words in Sentences subtest of the MLAT in their incidental classroom-context study, earlier training in grammar may bias the Words in Sentences subtest results in favor of the assumption that LAA would have an important role in late immersion students’ success.

Second, encoding the input data and the extraction of form-function relations in the context of the target lexical items requires inductive language ability (Williams, 1999). In addition, inferring the meanings of the lexical items from their repeated use in different contexts could call for inductive learning mechanisms (N. Ellis, 1994). Third, the PLAB makes an appropriate aptitude test for high school students (Skehan, 2002).

In the Language Analysis test, participants were presented with the words from a foreign language and the English equivalent of these words. Then, using multiple-choice items, they had to figure out the equivalent of 15 English statements in that language. Reliability of the test was calculated on the scores of the participants involved in the treatments (n = 65), using Cronbach’s alpha again. The obtained alpha was low, (α = 0.53). This can be related to the low variation in the scores (M = 11.90, SD = 2.13) (Erlam, 2005). The low number of items on the test compared to other tests used may also explain this low reliability (Fulcher, 2010).

c Working memory test

Operation span, counting span, and reading span tasks have been widely used in cognitive psychology as both valid and reliable measures of WMC (Conway et al., 2005). Turner and Engle (1989) hypothesized that the processing element of the span task and WMC are independent. For instance, it is reasonable to use a WM span test, which does not tap into sentence reading, for exploring reading ability. In the present study, WM was assessed by an operation span task following Turner and Engle (1989). Each operation string consisted of a simple arithmetic operation and its solution, with half of the answers correct and half of them incorrect. The operations contained a simple division or multiplication problem followed by a subtraction or addition of a number. Each operation was followed by a to-be-remembered word. Before administration, the WM test was piloted and the transition time was calculated at five seconds for each operation being presented on a PowerPoint slide.

The WM test was administered individually. The set size, or the number of operations, increased from two to five, and there were three trials at each set size. L2 learners had to read each operation and say aloud whether its answer was correct or incorrect. Then, immediately, they were to go to the next operation. At the end of each set, they were asked to repeat the words they had seen in front of the operations. If 85% of each individual’s answers to the operations were right, their scores on word recall were taken into account. With regard to the scoring procedure, and considering the suggestion of Conway et al. (2005), each correctly recalled word was awarded one point, and the sum of the points for all the trials was an individual’s span score. Reliability, using Cronbach’s alpha (n = 65), was α = .87 (M = 28.66, SD = 5.62).

d Receptive knowledge test

This test was created based on Webb, Newton, and Chang (2013) for measuring receptive knowledge of form and meaning, but it was slightly different to favor our implicit treatments (Sonbul & Schmitt, 2013) by asking L2 learners to engage in more contextualized L2 use. This test had 20 items and required L2 learners to write the L1 meanings of the target collocations:

I had to rustle up some food. We were hungry.

Rustle up means---------------------------------

The items were scored dichotomously: the correct items were awarded one point while the incorrect (e.g. L1 translations of ‘cook’, ‘make’, or ‘eat’ for the phrasal verb ‘rustle up’) or unanswered ones were given zero point. Since reliability depends on test length (Fulcher, 2010), reliability of the test, calculated on the sum of the scores on the posttest and delayed posttest (see Erlam, 2005), was α = .71 (M = 32.09, SD = 4.58).

e Productive knowledge test

This test was administered before the receptive knowledge test in order not to let L2 learners learn from the tests themselves. The test format was based on Forsberg Lundell and Sandgren (2013) to measure productive knowledge of form and meaning. The first letter of the dominant word of each collocation was granted in order to avoid alternative answers, and participants had to fill in the blanks with the collocations that fit the context:

The children and I were terribly hungry. We had to r _ _ _ _ _ _ _ some food.

The correct provision of the target collocations received a score of one; L2 learners had to write the correct form of the verb and preposition of each collocation to achieve the total score. The incorrect production of the target collocations received a score of zero. For instance, a correct verb with an incorrect preposition (e.g. rustle on) or an incorrect verb with a correct preposition (e.g. russle up) was not given credit. Using the aggregated test scores on the posttest and delayed posttest, reliability of the test was α = .74 (M = 8.66, SD = 5.06).

5 Procedure

The treatment consisted of 100 English passages that contained 20 target grammatical collocations. All genuine passages (estimated to be similar in terms of linguistic complexity) were selected from the fiction section of the COCA and were long enough (around 60 words each) to be meaningful. The COCA presents collocations and their frequency in spoken, fiction, magazine, newspaper, and academic sections. According to Webb et al. (2013), five or more encounters in written context may be necessary to learn L2 collocations incidentally. When L2 learners are exposed to abundant input, the meaningful context of input provision brings about an implicit condition of learning that may help higher aptitude learners focus on form (Skehan, 2015). L2 learners encountered each grammatical collocation five times in the five passages and five more times in the statements following them.

Because genuine texts are produced by and for native speakers, they are complex and may only fulfill advanced learners’ needs (Long, 2007). To improve their comprehensibility, but without simplifying them, they can be elaborated by adding redundancy and clarification. In the elaborated version, the sentences can be both lexically and structurally elaborated. Lexical elaboration is achieved by providing paraphrases, restatements, and synonyms of low-frequency lexical items. For provision of structural elaboration, the logical thematic relations are actualized by such devices as intersentential linkers, full noun-phrases, anaphoric references instead of cataphoric ones, and retrieving omitted elements (Kim, 2006). Since elaborated texts can sometimes increase linguistic complexity by extending text length, in the modified elaborated version, sentences of the elaborated version are split, and the target items are typographically enhanced to foster FonF (Long, 2015a). In the present study, genuine passages were used for the GI group, and then they were lexically and structurally elaborated for the EI group. Because the target collocations were unfamiliar to L2 learners, their meanings were directly provided in the elaborated passages, or the sentences that contained them were elaborated in such a way that they fully clarified the meaning of the collocations. Since typographical enhancement of multi-word expressions such as collocations has shown to promote incidental collocation learning (e.g. Choi, 2017), in the modified elaborated passages, the target collocations were bolded and underlined. Moreover, long sentences of the elaborated version were divided into shorter sentences (see Appendix 2).

VI Results

The primary goal of this study was to examine to what extent IDs in language aptitude modify the relationship between genuine, modified, and elaborated input and L2 learners’ acquisition of English grammatical collocations.

The aptitude scores of L2 learners involved in the treatments (n = 65) were first subjected to correlation analyses to ascertain that they tested different skills or aptitude components. The results demonstrated that, although they were positive, there were no significant correlations between these measures. The results are shown in Table 2.

Table 2.

Correlations between measures of language aptitude.

Measure	1	2	3
1. Working memory		.11	.20
2. Language analysis	.11		.16
3. Sound–symbol association	.20	.16

Note. p < .01, two-tailed.

In order to verify that there were no significant differences between the groups with regard to their performance on the aptitude tests, a one-way between-groups analysis of variance (ANOVA) was conducted. The results showed that the scores were not significantly different: the WM test, F(2, 62) = .04, p = .96; the Language Analysis test, F(2, 62) = .86, p = .42; the Sound–Symbol Association test, F(2, 62) = 1.41, p = .25. Descriptive statistics for the outcomes of the treatment and control groups on the measures of language aptitude and measures of receptive and productive knowledge are presented in Table 3 and Table 4, respectively.

Table 3.

Descriptive statistics for aptitude scores.

Aptitude	GI^a		EI^b		MEI^c
	M	SD	M	SD	M	SD
WM	28.27	5.37	28.59	4.71	28.14	6.35
LAA	12.36	1.94	12.00	2.02	11.52	2.33
PCA	18.45	3 .09	18.50	2.68	17.23	2.50

Note. GI = genuine input; EI = elaborated input; MEI = modified elaborated input; WM = working memory; LAA = language analytic ability; PCA = phonetic coding ability. ^a n = 22. ^b n = 22. ^c n = 21.

Table 4.

Descriptive statistics for tests of outcome.

Test	GI		EI		MEI		Control
	M	SD	M	SD	M	SD	M	SD
Productive knowledge:
Pretest	.00	.00	.00	.00	.00	.00	.00	.00
Posttest	4.50	2.34	4.45	2.32	5.85	2.28	.00	.00
Delayed posttest	3.45	2.80	3.90	2.04	5.19	2.52	.00	.00
Receptive knowledge:
Pretest	.00	.00	.00	.00	.00	.00	.00	.00
Posttest	14.36	2.19	16.27	1.51	17.52	1.16	.00	.00
Delayed posttest	12.36	1.36	16.86	1.42	18.57	1.07	.00	.00

Note. GI = genuine input; EI = elaborated input; MEI = modified elaborated input.

Toward the major goal of the present study, a series of Pearson r correlation coefficients were computed between the students’ scores on the posttest-delayed posttest and their scores on the measures of language aptitude. The results are displayed in Tables 5 -7. They show how aptitude scores as measured by the WM, Language Analysis, and Sound–Symbol Association tests are related to L2 learners’ performance on the posttest and delayed posttest of productive knowledge and receptive knowledge.

Table 5.

Correlations between aptitude scores and gain scores for the genuine input (GI) group.

Aptitude	Productive knowledge		Receptive knowledge
	Posttest	Delayed posttest	Posttest	Delayed posttest
WM	.54^**	.62^**	.50^**	.25
LAA	.24	.10	.04	.03
PCA	.10	.01	.14	.15

Note. WM = working memory; LAA = language analytic ability; PCA = phonetic coding ability. **p < .01, two-tailed.

Table 6.

Correlations between aptitude scores and gain scores for the elaborated input (EI) group.

Aptitude	Productive knowledge		Receptive knowledge
	Posttest	Delayed posttest	Posttest	Delayed posttest
WM	.58^**	.65^**	.47^**	.19
LAA	.30	.36	.34	.01
PCA	.36	.26	.42	.09

Note. WM = working memory; LAA = language analytic ability; PCA = phonetic coding ability. **p < .01, two-tailed.

Table 7.

Correlations between aptitude scores and gain scores for the modified elaborated input (MEI) group.

Aptitude	Productive knowledge		Receptive knowledge
	Posttest	Delayed posttest	Posttest	Delayed posttest
WM	.50^**	.58^**	.11	.12
LA	.17	.18	.11	.05
PCA	.09	.01	.21	.24

Note. WM = working memory; LAA = language analytic ability; PCA = phonetic coding ability. **p < .01, two-tailed.

In the GI group, there was a significant correlation between WM and the posttest and delayed posttest of productive knowledge. In addition, the scores on the WM test correlated significantly with the posttest scores of receptive knowledge. In this group, the LAA and PCA measures were not strongly related to the learning achievements. In the EI group, significant correlations were obtained between similar variables, with high levels of WM associated with high gain scores on the posttest and delayed posttest of productive knowledge and the posttest of receptive knowledge. Results for the MEI group, using Pearson’s product moment correlations, revealed that WM correlated significantly with L2 learners’ performance on the posttest and delayed posttest of productive knowledge. There were no significant correlations between the measures of language aptitude and short- and long-term gains in receptive knowledge.

In order to explore the interrelationship between the significantly correlated variables, that is, to investigate how well WM could account for the variance in gain scores, simple linear regression analyses were performed. Before regression analyses, preliminary analyses were conducted to ensure there were no violations of the assumptions of normality, linearity, and homoscedasticity. The results of the linear regressions are presented in Table 8. Considering the correlations between WM and productive knowledge in the treatment groups, regression analyses indicated that about 26% of the variance in the posttest of productive knowledge and 33% of the variance in the delayed posttest of productive knowledge (dependent variables) could be explained by WM (independent variable).

Table 8.

Results from simple linear regression analyses.

	Productive knowledge						Receptive knowledge
	Posttest			Delayed posttest			Posttest
Predictor	B	β	R²	B	β	R²	B	β	R²
WM	.22	.51	.26^*	.27	.57	.33^*	.19	.45	.21^*

Note. WM = working memory. *p < .05.

In the GI and EI groups, WM correlated significantly with the posttest of receptive knowledge. Taking the scores of these groups, a simple linear regression was carried out. The results revealed that the WM scores significantly predicted the posttest scores. Indeed, WM explained 21% of the variance in the posttest scores. The B weights indicated that WMC made a greater contribution to productive knowledge (B = .22 and B = .27) than to receptive knowledge (B = .19).

Although the focus of the current article, as part of a larger project, is not on the relative effectiveness of the different types of input, we will briefly present the differences among the outcomes of the treatment conditions. Analysis of the pretest showed zero variance, indicating that L2 learners were not familiar with the target collocations before the treatment. An ANOVA was conducted to investigate the relative effectiveness of genuine input, elaborated input, and modified elaborated input on L2 learners’ performance on the tests. Receptive and productive test scores were normally distributed in all three treatment groups, and no outliers were detected in the data. The alpha level of .05 was set for making decisions. The results revealed that there was no significant difference between the groups with regard to their posttest and delayed posttest of productive knowledge. As for the receptive knowledge scores, the three groups were significantly different from one another on the posttest and delayed posttest. Using the Tukey HSD test, post-hoc comparisons for the posttest of receptive knowledge indicated that the EI group performed significantly better than the GI group. In addition, the MEI group achieved significantly higher learning gains than both the GI and EI groups. Post-hoc comparisons for the delayed posttest of receptive knowledge revealed that both the EI and MEI groups outperformed the GI group significantly. Moreover, The MEI group recalled significantly more receptive collocation knowledge than the EI group.

VII Discussion

The present study was designed to investigate the contribution of language aptitude components to the learning of grammatical collocations under three input conditions: genuine input, elaborated input, and modified elaborated input. It was initially assumed that there might be positive correlations between the main aptitude components in Skehan’s (2002, 2012) Processing Stage model and L2 learners’ achievements. Indeed, LAA and PCA correlated with learning gains in all the treatment groups, but not significantly. The only significant correlations were observed for WM. This suggests that Robinson (2012) may have rightly stated that combinations of cognitive abilities are differentially connected with processing under different pedagogical conditions of receiving L2 input. Among the four communicative-FonF conditions that Robinson specified, it was acquisition from an implicit condition (via written input) that is associated with WMC.

Concerning our first research question, in the GI group, WM correlated significantly with productive knowledge scores on the posttest and delayed posttest, and with receptive knowledge scores on the posttest. The input that L2 learners received was beneficial to those who had higher WMC; they comprehended, and produced the target items better than the other L2 learners in this group, probably because their WM resources helped to direct their attention to relevant linguistic features in input, retaining linguistic units in memory for additional processing, and removing unnecessary information (Kormos, 2013). The results are in accordance with N. Ellis’s (1996) statement that WM influences collocation learning in implicit conditions, within various input modalities, including written input.

In terms of receptive knowledge, no significant correlations were observed between WM and the scores on the delayed posttest. According to DeKeyser and Koeth (2011), high WM learners may be able to glean more information to process than low WM learners. The larger amount of data may entail processing and consolidating during a longer period of time. This is a possible reason that WMC became predictive of the delayed gains in some studies (e.g. Erlam, 2005; Li, 2013). On the contrary, in the present study, genuine input, which provided repeated exposure to the target collocations, helped both high and low WM learners to pick up and hold data to process over time, as evidenced by a lack of relationship between WM and long-term receptive knowledge of collocations.

The second research question involved exploration of the relationship between language aptitude and learning grammatical collocations in the EI group. Comparable to the GI group, LAA and PCA did not correlate significantly with any learning scores, but WM correlated significantly with the posttest and delayed posttest of productive knowledge and the posttest of receptive knowledge. That is, L2 learners with higher WMC were probably better able to direct their attention to the target items (Kane et al., 2001) and achieve better results.

The EI group’s performance on the posttest and delayed posttest of receptive knowledge was significantly superior to that in the GI group. This indicates that Long and Ross (2009) rightly expressed that elaborated input, despite its potential side effects such as longer and more linguistically complex texts than the original, yields better comprehension than genuine input. With regard to the IDs, similar to the genuine input condition, the significant relationship between WM and the receptive knowledge posttest suggested that WM was a predictor of success under the elaborated input condition, given that L2 learners with higher WMC were better able to gain the knowledge of collocational forms and meanings. A standard multiple regression (for the GI and the EI groups) confirmed that WM accounted for 20% of the variance in the receptive knowledge posttest scores. The findings lend support to Masson and Miller’s (1983) perception of WM because they reveal that WM is important for attending to the meaning of unknown words implicitly stated in a passage. Concerning the delayed posttest, Pearson correlation revealed that the delayed posttest scores were not significantly related to the WM scores. The results suggest that the elaborated input condition eased both high and low WM students’ consumption of their WM resources for holding information and processing it over time.

The productive knowledge posttest and delayed posttest results were nearly the same in the EI group and GI group. Moreover, the correlation found between WM and productive knowledge in the EI group was significant and closely approximated the correlation between the same variables in the GI group. Despite the fact that elaborated input improves comprehensibility, it can make texts longer and more complex than genuine texts. It has been suggested that L2 learners’ attentional resources are not boundless, and they may give priority to meaning rather than form within time limits (Skehan, 2009). The complex elaborated texts probably made the EI group focus more on meaning rather than form under time pressure (it has already been mentioned that the time was the same for all the groups), therefore, this group was not able to produce the target collocations better than the GI group.

In terms of the association between language aptitude and modified elaborated input condition, as investigated in the third research question, the results indicated that neither the posttest nor the delayed posttest of receptive knowledge correlated significantly with the measures of language aptitude. Furthermore, the receptive knowledge scores obtained by those L2 learners who received modified elaborated input became significantly better than the scores of the GI and EI groups. Enhanced target collocations in modified elaborated input provided L2 learners with the opportunity to engage in FonF, and shorter sentences in this type of input made it less complex compared to elaborated input. This can have directed L2 learners’ attention to both meaning and form, compatible with Doughty’s (2001) affirmation that selective attention to features of learning items promote processing for meaning (as the L2 learners’ default mode) and encoding the forms. Indeed, focusing on form while L2 learners are mainly focused on meaning assists form-function mapping.

Although, concerning the results of the GI and EI groups, it was likely that WM scores would correlate with the scores on the receptive knowledge posttest, easier processing might have increased the amount of information reserved and processed in WM (Swanson & Berninger, 1996). In this way, both high and low WM learners benefited to similar degrees from modified elaborated input. The findings provide evidence for the value of modified elaborated input, which promotes attention to form and meaning (Long, 2015a), and reduces demands on WM by alleviating identification, selection, and association of critical features in input (Martin & N. Ellis, 2012).

The scores on the posttest and delayed posttest of productive knowledge improved in comparison with the other groups, although the differences were not significant. This improvement can pertain to the notable features of modified elaborated input such as less complexity and more FonF. However, similar to the GI and EI groups, the correlation between WM and the posttest and delayed posttest of productive knowledge was significant in the MEI group. Although, under modified elaborated input condition, L2 learners received less complex texts with enhanced collocations, processing the collocational forms within their contexts demanded high WM resources. These findings suggest that WMC is an ability relevant to explain productive knowledge of grammatical collocations, and consistent with previous research (e.g. Swanson & Berninger, 1996), the cognitively demanding processes in language production are closely associated with WMC. Furthermore, according to the transfer-appropriate processing framework, the nature of a learning task will determine the nature of the resulting knowledge (e.g. Barcroft, 2004). Given that reading tasks as such do not involve language production, it is to be expected that they foster receptive knowledge more than productive knowledge. Modified elaborated input may have eased WM processes more for gaining receptive knowledge than productive knowledge. The results contrast with those of Indrarathne and Kormos (2017) who found that WM had a stronger link to receptive knowledge than to productive knowledge in implicit contexts of receiving written input. This contrast may be due to the different conditions of taking the productive knowledge test by L2 learners since in Indrarathne and Kormos’s study, productive knowledge was not tested under time pressure.

Considering WM and the productive knowledge scores of all the groups, WM was observed to be a significant predictor of productive knowledge in the posttest (R² = 26%) and delayed posttest (R² = 33%) when entered into the standard multiple regression analysis. That is to say, L2 learners with high WMC produced L2 collocations better than individuals with limited WMC.

VIII Conclusions

This study explored the role of language aptitude in acquiring English grammatical collocations embedded within genuine input, elaborated input, and modified elaborated input. Four major findings emerged: first, WM seems to play a greater role than PCA and LAA for implicit learning of grammatical collocations through written input that provides FonF. Second, if L2 learners are presented with modified elaborated input, WM will no longer be a strong predictor of success in a receptive knowledge test. This type of input that provides more opportunities for FonF may make L2 input remain active in WM and ‘levels the playing field’ (Vatz et al., 2013, p. 289) for both low and high WM students to increase their receptive knowledge of new grammatical collocations. Third, long-term receptive knowledge of L2 grammatical collocations can be fostered by repeated exposure in genuine, elaborated, and modified elaborated input. This flooding component allows for noticing collocations and their formal and morphological features in L2 input and may help L2 learners pick up and keep more data in their WM for gradual, incremental learning. Fourth, WM is heavily involved in gaining productive knowledge of grammatical collocations from the kinds of reading input used in this study.

Limitations of the present study as well as suggestions for further research need to be underscored. To begin with, one of the limitations of this study pertains to its generalizability. Participants were intermediate-proficiency students and participated in a program for learning L2 grammatical collocations. Future research is needed to uncover the role of language aptitude in L2 learners with different levels of proficiency in being able to learn different types of L2 collocations under input elaboration conditions. Furthermore, future studies can be carried out to investigate the role of deep semantic processing, as the other hypothesized cognitive factor involved in implicit-FonF condition of presenting written input (Robinson, 2012), in learning L2 collocations. Another limitation of the present study is that only the written mode was used at both the input and test stages. Listening input and oral test measures could help to further examine the link between various components of aptitude and the acquisition of L2 collocations. As another limitation, it is possible that the lack of a significant relationship between LAA and L2 learners’ receptive and productive knowledge was a side effect of the low reliability of the measure of LAA (α = 0.53). Replication studies might report higher reliability of a measure of LAA and contradict or confirm the findings of the present research. Finally, none of the treatments were more favorable in terms of productive knowledge, and only high WM learners in the treatment groups were successful in producing the collocations. In future research, it will be important to provide L2 learners with more collocation instances to see which type of input may better reveal the impact of frequency of exposure on productive knowledge and may neutralize the effects of WM.

According to Vatz et al. (2013), the practical implications of the results in ATI studies depend on identifying the interaction between language aptitude and treatments as well as the relative effects of the treatments. The current study expands our understanding of the contribution of cognitive differences to intermediate adult L2 learners’ collocational knowledge under implicit-FonF conditions of presenting written input and has important pedagogical implications for L2 teaching. To begin with, the investigation of the significant group effects and the finding that WM component of language aptitude and learning collocations under elaborated input condition are closely intertwined suggest that for a given individual with a high score on a WM measure, a flood of grammatical collocations embedded in elaborated input emerges as a fruitful approach to generate immediate and long-term gains in receptive collocation knowledge. Second, in order to optimize learning for a group of learners that is (almost inevitably) heterogeneous in terms of WMC, teachers can adopt the use of modified elaborated input to neutralize the within-group differences in WMC, and thus to help more L2 learners increase immediate and long-term receptive knowledge of grammatical collocations.

Footnotes

Appendix 1. The 20 grammatical collocations that were the target items to learn

Collocation	Meaning	Example
1. blaze away	To fire bullets rapidly and continuously	They blazed away at a beautiful herd of elk.
2. bolt down	To eat very quickly	She bolted down her lunch in five minutes.
3. brim over	To be full of something	She was brimming over with happiness.
4. chuck out	To throw something away	Just go ahead and chuck out the batteries.
5. fork out	To spend a lot of money on something	They had to fork out for a new car.
6. gasp out	To say something while you are breathing with difficulty	She gasped out a few words haltingly.
7. grope about	To search for something by feeling with your hands	He groped about for his glasses.
8. jabber away	To talk rapidly and unintelligibly	The friend jabbered away for hours.
9. limber up	To do gentle exercises; to warm up	The athletes are limbering up.
10. loll around	To sit or lie in a relaxed way	The tired travelers lolled around all over the hotel lobby.
11. muck up	To spoil something	The bad weather mucked up our picnic plans.
12. palm off	To trick someone to buy or take something	They tried to palm off their old sofa on us.
13. pelt down	To rain heavily	We can’t go out because it’s pelting down.
14. rap out	To say something loudly and quickly	The major rapped out his orders.
15. rustle up	To make something quickly, especially a meal	He rustled up breakfast.
16. salt away	To put aside; save	She salted away some money for her education.
17. smarten up	To make something look better	They decided to smarten up the house.
18. tone up	To improve the strength of your body	Aerobics tones up your muscles.
19. trail away	To become quieter or weaker	His voice trailed away.
20. trifle with	To treat someone or something without respect	Don’t trifle with me.

Appendix 2. A grammatical collocation within genuine,elaborated,and modified elaborated input

Acknowledgements

We would like to thank Professor Frank Boers, the co-editor of Language Teaching Research at the time of reviewing this manuscript, anonymous reviewers, and Dr. Mohammad Javad Ahmadian for their insightful suggestions and comments. We would also like to thank the students who kindly helped us with data collection with their cooperation, punctuality and commitment.

Conflict of Interest

This study is part of a larger study concerned with both the interaction of language aptitude with different input learning conditions and the relative effects of these input conditions without considering the effect of language aptitude. The second part of this lager research has been briefly addressed in the current study (see Table 4) to generate a more extensive research discussion and to provide readers with a clearer picture of the results. This part, however, has been fully discussed in another manuscript, which has been accepted for publication (Farshi, Tavakoli, & Ketabi, in press).

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Baddeley

A.D.

(2000). The episodic buffer: A new component of working memory? Trends in Cognitive Science, 4, 417–423.

Barcroft

(2004). Second language vocabulary acquisition: A lexical input processing approach. Foreign Language Annals, 37, 200–208.

Barfield

(2013). Lexical collocations. In Chapelle

C.A.

(Ed.), The encyclopedia of applied linguistics (pp. 3637–3641). Chichester: Wiley-Blackwell.

Carroll

J.B.

(1981). Twenty-five years of research on foreign language aptitude. In Diller

K.C.

(Ed.), Individual differences and universals in language learning aptitude (pp. 119–154). Rowley, MA: Newbury House.

Carroll

J.B.

Sapon

S.M.

(1959). Modern languages aptitude test. New York: Psychological Corporation.

Choi

(2017). Processing and learning of enhanced English collocations: An eye movement study. Language Teaching Research, 21, 403–426.

Conway

A.R.

Kane

M.J.

Bunting

M.F.

Hambrick

D.Z.

Wilhelm

Engle

R.W.

(2005). Working memory span tasks: A methodological review and user’s guide. Psychonomic Bulletin & Review, 12, 769–786.

DeKeyser

Koeth

(2011). Cognitive aptitudes for L2 learning. In Hinkel

(Ed.), Handbook of research in second language teaching and learning: Volume II (pp. 395–406). London: Routledge.

Doughty

C.J.

(2001). Cognitive underpinnings of focus on form. In Robinson

(Ed.), Cognition and second language instruction (pp. 206–257). Cambridge: Cambridge University Press.

10.

Ellis

N.C.

(1994). Vocabulary acquisition: The implicit ins and outs of explicit cognitive mediation. In Ellis

(Ed.), Implicit and explicit learning of languages (pp. 211–282). London: Academic Press.

11.

Ellis

N.C.

(1996). Sequencing in SLA: phonological memory, chunking and points of order. Studies in Second Language Acquisition, 18, 91–126.

12.

Ellis

(2012). Language teaching research and language pedagogy. Chichester: Wiley-Blackwell.

13.

Erlam

(2005). Language aptitude and its relationship to instructional effectiveness in second language acquisition. Language Teaching Research, 9, 147–171.

14.

Farshi

Tavakoli

Ketabi

(in press). Receiving Different Types of Input for Learning and Retention of English Grammatical Collocations in EFL Classrooms. Pedagogies: An International Journal.

15.

Forsberg Lundell

Sandgren

(2013). High-level proficiency in late L2 acquisition: Relationships between collocational production, language aptitude and personality. In Granena

Long

M.H.

(Eds), Sensitive periods, language aptitude, and ultimate L2 attainment (pp. 231–258). Amsterdam/Philadelphia, PA: John Benjamins.

16.

Fulcher

(2010). Practical language testing. London: Hodder Education.

17.

Grabe

(2009). Reading in a second language: Moving from theory to practice. New York: Cambridge University Press.

18.

Granena

(2013). Individual differences in sequence learning ability and second language acquisition in early childhood and adulthood. Language Learning, 63, 665–703.

19.

Granena

Long

M.H.

(2013). Age of onset, length of residence, language aptitude, and ultimate L2 attainment in three linguistic domains. Second Language Research, 29, 311–343.

20.

Harley

Hart

(1997). Language aptitude and second language proficiency in classroom learners of different starting ages. Studies in second language acquisition, 19, 379–400.

21.

Indrarathne

Kormos

(2017). The role of working memory in processing L2 input: Insights from eye-tracking. Bilingualism: Language and Cognition, 21, 355–374.

22.

Juffs

Harrington

(2011). Aspects of working memory in L2 learning. Language Teaching, 44, 137–166.

23.

Kane

M.J.

Bleckley

M.K.

Conway

A.R.

Engle

R.W.

(2001). A controlled-attention view of working-memory capacity. Journal of Experimental Psychology: General, 130, 169–183.

24.

Kim

(2006). Effects of input elaboration on vocabulary acquisition through reading by Korean learners of English as a foreign language. TESOL Quarterly, 40, 341–373.

25.

Kormos

(2013). New conceptualizations of language aptitude in second language attainment. In Granena

Long

M.H.

(Eds), Sensitive periods, language aptitude, and ultimate L2 attainment (pp. 131–152). Amsterdam/Philadelphia, PA: John Benjamins.

26.

(2013). The interactions between the effects of implicit and explicit feedback and individual differences in language analytic ability and working memory. The Modern Language Journal, 97, 634–654.

27.

Long

M.H.

(2007). Texts, tasks, and the advanced learner. In Long

M.H.

(Ed.), Problems in SLA (pp. 119–138). Mahwah, NJ: Erlbaum.

28.

Long

M.H.

(2015a). Second language acquisition and task-based language teaching. Malden, MA: Wiley-Blackwell.

29.

Long

M.H.

(2015b). Experimental perspectives on classroom interaction. In Markee

(Ed.), The handbook of classroom discourse and interaction (pp. 60–73). Malden, MA: Wiley-Blackwell.

30.

Long

M.H.

Ross

(2009). Input elaboration: A viable alternative to ‘authentic’ and simplified texts. In Namai

Fukada

(Eds), Toward the fusion of language, culture and education: From the perspectives of international and interdisciplinary research: A festschrift for Yasukata Yano (pp. 307–325). Tokyo: Kaitakusha.

31.

Martin

K. I.

Ellis

(2012). The roles of phonological short-term memory and working memory in L2 grammar and vocabulary learning. Studies in Second Language Acquisition, 34, 379–413.

32.

Masson

M.E.

Miller

J.A.

(1983). Working memory and individual differences in comprehension and memory of text. Journal of Educational Psychology, 75, 314–318.

33.

Meara

(2005). LLAMA language aptitude tests: The manual. Swansea: Lognostics.

34.

Nagata

Aline

Ellis

(1999). Modified input, language aptitude and the acquisition of word meanings. In Ellis

(Ed.), Learning a second language through interaction (133–149). Amsterdam/Philadelphia, PA: John Benjamins.

35.

Pimsleur

(1966). Pimsleur Language Aptitude Battery (PLAB). New York: The Psychological Corporation.

36.

Polio

Fleck

Leder

(1998). ‘If I only had more time’: ESL learners’ changes in linguistic accuracy on essay revisions. Journal of Second Language Writing, 7, 43–68.

37.

Robinson

(2002). Effects of individual differences in intelligence, aptitude and working memory on adult incidental SLA. In Robinson

(Ed.), Individual differences and instructed language learning (pp. 211–266). Amsterdam/Philadelphia, PA: John Benjamins.

38.

Robinson

(2012). Individual differences, aptitude complexes, SLA processes, and aptitude test development. In Pawlak

(Ed.), New perspectives on individual differences in language learning and teaching (pp. 57–75). New York: Springer.

39.

Sáfár

Kormos

(2008). Revisiting problems with foreign language aptitude. IRAL-International Review of Applied Linguistics in Language Teaching, 46, 113–136.

40.

Sheen

(2007). The effect of focused written corrective feedback and language aptitude on ESL learners’ acquisition of articles. TESOL Quarterly, 41, 255–283.

41.

Skehan

(1998). A cognitive approach to language learning. Oxford: Oxford University Press.

42.

Skehan

(2002). Theorizing and updating aptitude. In Robinson

(Ed.), Individual differences and instructed language learning (pp. 69–94). Amsterdam/Philadelphia, PA: John Benjamins.

43.

Skehan

(2009). Modelling second language performance: Integrating complexity, accuracy, fluency, and lexis. Applied Linguistics, 30, 510–532.

44.

Skehan

(2012). Language aptitude. In Gass

Mackey

(Eds), The Routledge handbook of second language acquisition (pp. 381–395). London: Routledge.

45.

Skehan

(2015). Foreign language aptitude and its relationship with grammar: A critical overview. Applied Linguistics, 36, 367–384.

46.

Sonbul

Schmitt

(2013). Explicit and implicit lexical knowledge: Acquisition of collocations under different input conditions. Language Learning, 63, 121–159.

47.

Swanson

H.L.

Berninger

V.W.

(1996). Individual differences in children’s working memory and writing skill. Journal of Experimental Child Psychology, 63, 358–385.

48.

Turner

M.L.

Engle

R.W.

(1989). Is working memory capacity task dependent? Journal of Memory and Language, 28, 127–154.

49.

Vatz

Tare

Jackson

S.R.

Doughty

C.J.

(2013). Aptitude–treatment interaction studies in second language acquisition: Findings and methodology. In Granena

Long

M.H.

(Eds), Sensitive periods, language aptitude, and ultimate L2 attainment (pp. 272–292). Amsterdam/Philadelphia, PA: John Benjamins.

50.

Webb

Newton

Chang

(2013). Incidental learning of collocation. Language Learning, 63, 91–120.

51.

Wen

(2015). Working memory in second language acquisition and processing: The phonological/executive model. In Wen

Mota

M.B.

McNeill

(Eds), Working memory in second language acquisition and processing (pp. 41–62). Bristol: Multilingual Matters.

52.

Williams

J.N.

(1999). Memory, attention, and inductive learning. Studies in Second Language Acquisition, 21, 1–48.

53.

Yalçın

Ş.

Spada

(2016). Language aptitude and grammatical difficulty. Studies in Second Language Acquisition, 38, 239–263.

54.

Yilmaz

Granena

(2016). The role of cognitive aptitudes for explicit language learning in the relative effects of explicit and implicit feedback. Bilingualism: Language and Cognition, 19, 147–161.