2 Theoretical Background

The role of vocabulary, as one of the subsystems of language, has often been downplayed in language teaching approaches during recent decades (Nunan, Reference Nunan2003). Traditionally, language was mainly assumed as a syntactic rule-governed process with little or no attention to the role of lexical items (Moudraia, Reference Moudraia2001). This refers back to the era when structural linguistics and audiolingualism were so popular. Advocates of audiolingualism believed that learners’ efforts must mainly focus on mastering grammar. In fact, they downgraded lexis as the basis of language, and held that once basic structures of a language had been mastered, new vocabulary could simply be ‘slotted in’ (Nunan, Reference Nunan2003). In other words, students were supposed to learn a lot of words, usually nouns to name a lot of things, and then use syntax to talk about those things. They were expected to use sentence grammar to do what they had never been prepared for, that is, linguistic novelty and creativity (Lewis, Reference Lewis2000, Reference Lewis2008).

In the twenty-first century, many teachers and educational systems still adhere to these stereotypes and view language as a composite of discrete components to be learned and practiced independently. Further, most students in these contexts come to associate this methodology with language classes despite the fact that their completion of such (structural) exercises generally does not enable them to express themselves creatively in the L2. Recently, Lewis (Reference Lewis1993, Reference Lewis2000, Reference Lewis2008) has taken this traditional view as a point of departure for his Lexical Approach and argues that this lexicalized-grammar view (i.e., “fixed” vocabulary and “generative” grammar) is an “invalid over-simplification” (2008: 11). He further contends that emphasis on mere sentence grammar (rather than word or text grammar) only provides the rules to analyze language but not ready-made lexicons as mental storage to which grammatical knowledge can be applied in novel language use. He then submits that once the learner has stored sufficient fixed and semi-fixed prefabricated chunks or lexis in the mental lexicon, s/he can recall and map them onto grammatical structures for any future linguistic novelty and creativity. Therefore, learners can best learn language when they use such components to communicate meanings. The Lexical Approach thus places communication of meaning at the heart of language and language learning and attaches more importance to the main carrier of meaning, i.e., vocabulary or lexis (Lewis, Reference Lewis2008).

This perspective, as Lewis (Reference Lewis2008) notes, is basically inspired by the Natural Approach and subsequently invigorated by emergentist, usage-based and dynamic systems theories (Dörnyei, Reference Dörnyei2009; Weinert, Reference Weinert2010). These theories hold the basic premise that language development is a self-organizing, complex, and dynamic phenomenon, grounded in lower, emergent processes (Dörnyei, Reference Dörnyei2009). From an emergentist perspective, although most structural syllabuses have presumed that teaching should be organized, linear and systematic, learning is, in effect, non-linear, holistic, cyclical, and evolves over time. In other words, teachers cannot control what students learn, in what order they learn, or how fast they learn (Willis, Reference Willis1990). Learning is not simply an additive process, and it involves learners re-organizing their previous interlanguage. This is because new input requires learners constantly to make re-adjustments to what they have internalized. For instance, learners may not get the present perfect before they have understood the simple present and the simple past tenses, and this reflects the interconnections of all verb forms in relation to each other. That is, tenses are interrelated and the learners’ understanding of a particular verb form develops as they face different applications of a verb tense and experience its uses in relation to other tenses (Lewis, Reference Lewis2000).

Hopper (Reference Hopper2002) defines this cyclical process in which movement towards a complete construction is constant while completion is always deferred as ‘emergent grammar’. Emergent grammar is a set of routinized and frequently used constructions, accumulated fragment by fragment as learners become more and more involved in communicative activities (Hopper, Reference Hopper1998). The creative production of these routinized constructions demands prior knowledge of prefabricated chunks, available for use in appropriate contexts. Based on this view, the rules of the system are intrinsically incomplete and evolve over time out of exemplars of simple, local schemas (Hopper, Reference Hopper2002).

This view, as noted earlier, lies at the heart of Lewis’s account of grammaticalized lexis which, in turn, has an important implication for English language teaching (ELT). In the Lexical Approach, the central theme is “raising students’ awareness of, and developing their ability to ‘chunk’ language successfully” (Lewis, Reference Lewis1993: vi). The notion of chunking relates to the ways lexical items are naturally stored and retrieved in the memory as wholes (Lewis, Reference Lewis1993). From a psycholinguistic vantage point, the basis of native speaker fluency is control of a vast repertoire of these chunks and formulaic lexical phrases (or formulaicity). Also, research on first language (L1) acquisition (e.g., Tomasello, Reference Tomasello2000) has shown that exposure to highly repetitive chunks or lexical frames learned as wholes is an intrinsic part of L1 acquisition, and, insofar as L2 attainment involves similar (linguistic or learning) mechanisms and human learners, certain commonalities in terms of underlying psychological processes are highly likely. For instance, MacWhinney (Reference MacWhinney2004) and Dörnyei (Reference Dörnyei2009) argue that the key learning processes underlying child L1 acquisition (e.g., statistical and frequency-based learning) are observable in SLA. The Lexical Approach thus assumes that the processes involved in the learning of L1 and L2 can be similar in that the lexical phrases or unanalyzed wholes play a crucial role in both processes (Hunston & Francis, Reference Hunston and Francis2000). Recent interest in chunking and formulaicity has even aspired Kweldju (Reference Kweldju2004) to look for neurolinguistic evidence to support the notion. He argues that the right hemisphere is initially involved in processing and storing formulaic expressions or unanalyzed wholes, with the left hemisphere subsequently coming into play to analyze the language patterns for future creative production. As the pioneer of this notion, Seliger (Reference Seliger1982) notes that the data first processed in the right hemisphere can be further processed in the left hemisphere for linguistic creativity. This neurolinguistic account in turn elucidates why drills do not seem to contribute to immediate natural language use. To put it another way, the basic concern to the Lexical Approach is how L2 learners can build mental phrasal lexicons which are similar to those created by native speakers (Lewis, Reference Lewis2000).

Other SLA researchers such as Ellis and Larsen-Freeman (Reference Ellis and Larsen-Freeman2006) seem to be intrigued by usage-based premises similar to those inspiring a lexis-based view of L2 learning. They use the term “grammaticalization” to refer to the emergence of syntactic constructions from frequent occurrence of linguistic elements for functional language use. In a similar vein, Diessel (Reference Diessel2004) argues that complex syntactic constructions may develop out of exemplars of simple ones. In the Lexical Approach, therefore, lexis has gained ground as an important aspect of language teaching and learning in L2 classrooms. In the lexis-based view, according to Lewis (Reference Lewis2008), a “good vocabulary” means not only knowledge of enough words but also an adequate phrasal lexicon. In this view, vocabulary no longer takes care of itself, as was held in traditional grammar-based views, but essential lexis should be selected and deliberately included in instructional materials and activities. The classroom schedule, besides ensuring sufficient exposure to lexis in context, should also aim to maximize the likelihood of learners turning lexical input into intake (Lewis, Reference Lewis2008). However, inclusion of an adequate lexical syllabus does not imply a one-brick-at-a-time approach to vocabulary input as the mental lexicon develops not linearly, but holistically. This view, by implication, assigns little control to teachers over what is actually learned because students learn a great deal from indirect (or incidental) exposure to language through reading, listening, or, in general, working with lexis, without the need for the teacher to present a description of what is taught (Willis, Reference Willis1990).

Inspired by the Lexical Approach, several studies (e.g., Channell, Reference Channell1994; Gilquin, Granger & Paquot, Reference Gilquin, Granger and Paquot2007; Koosha & Jafarpour, Reference Koosha and Jafarpour2006; Lee, Reference Lee2011; Park, Reference Park2012; Quinn, Reference Quinn2015; Rahimi & Momeni, Reference Rahimi and Momeni2012; Rahimi, Momeni & Nejati, Reference Rahimi, Momeni and Nejati2012) have investigated the effects of implementing a lexis-based view using concordancing materials on L2 learning. Their results point to the effectiveness of corpora in improving students’ L2 skills. Rahimi et al. (Reference Rahimi, Momeni and Nejati2012), for instance, found that concordancing packages had a positive effect on improving students’ achievement in vocabulary and reading. Similarly, Quinn (Reference Quinn2015) reported that corpora have the potential to support the L2 writing process at the discourse level and learner concordancing can offer students an alternative reference to traditional dictionary searches. Further evidence comes from Huang (Reference Huang2007) and Gardner (Reference Gardner2007) suggesting that learners need a great deal of exposure to reading materials emphasizing new lexis in order to better retrieve L2 vocabulary in their productive skills. However, other studies (e.g., McEnery, Wilson & Baker, Reference McEnery, Wilson and Baker1997; Schmidt, Reference Schmidt2001) have raised concerns about the presumed success of the Lexical Approach in grammar learning. Schmidt (Reference Schmidt2001), for example, argues that some type of explicit focus on form is also required for learners to notice and learn L2 grammatical structures. In a similar vein, McEnery et al. (Reference McEnery, Wilson and Baker1997), using corpora for grammar teaching at the pre-tertiary level in the UK, conclude that although integrating a corpus-based methodology is beneficial to improving language proficiency, it might be insufficient for developing higher levels of linguistic accuracy. Further, Sinclair (Reference Sinclair1991) suggests that consciousness-raising about the links between form and meaning might minimize the learning load, and that learners can use corpus evidence to develop individual creativity in language classrooms.

In recent years, advances in computer technology have contributed a great deal to vocabulary research and pedagogy (Nunan, Reference Nunan2003). One of these advances is the advent of corpus linguistics which has in turn led to the design of various databases or corpora, on- or offline. This modern linguistic trend is interested in probing language use in real-life situations and encourages inductive generalizations about emergent tendencies, or lexical co-occurrence and recurrence. Word meanings, in this view, are captured through searching for statistical tendencies of phraseology, collocations, colligations, cognates, and lexical bundles as well as their distributions in natural language data. Corpus linguistics has rather diminished the importance of traditional lexicology, or semantics of word meanings in isolation based on language-internal relations (Rajagopalan, Reference Rajagopalan2004). This new development could have a great impact on language teaching, especially from a lexical perspective. According to Lewis (Reference Lewis1993), evidence from corpora, computational linguistics, and discourse analysis should now influence syllabus content and sequencing. There are also corpus-oriented instructional materials which emphasize empirical data and present a better quality of learner input and a better understanding of lexical patterning in specific contexts (Campoy-Cubillo, Bellés-Fortuño & Gea-Valor, Reference Campoy-Cubillo, Bellés-Fortuño and Gea-Valor2010). Because of the accessibility to natural written and spoken corpora, students now have more opportunities to develop a better understanding of real language use than before (Lewis, Reference Lewis2000).

Three important UK-based corpora include the Collins Birmingham University International Language Database (COBUILD), the Cambridge International Corpus (CIC), and the British National Corpus (BNC) (cf. Richards & Rogers, Reference Richards and Rogers2001). The COBUILD Bank of English was the starting-point in this area under the direction of John Sinclair (O’Keeffe, McCarthy & Carter, Reference O’Keeffe, McCarthy and Carter2007). The purpose of this project was to provide learners with dictionaries and materials which focused on real language use and meanings which learners were often required to use in real communicative settings. The COBUILD learners’ dictionaries, grammars and usage manuals facilitate the incorporation into syllabuses of findings on frequency distribution and collocations through genuine rather than invented or intuitive examples (Römer, Reference Römer2010). Another corpus resource is the CIC which comprises vast, methodical collections of both spoken and written language. The CIC consists of more than one billion words, which along with COBUILD is constantly being updated to allow the monitoring of language usage (O’Keeffe et al., Reference O’Keeffe, McCarthy and Carter2007). The CIC also features learner corpora which include more than 27 million words of learners’ writing, 12 million of them error coded. Consequently, this provides information about the common lexical and grammatical errors which are made by learners and allows dictionary and materials writers to highlight common problems. Finally, the BNC is a freely available corpus that notably contains samples of spoken language (ecologically collected from real spontaneous discourse) as well as written language (obtained from a wide range of disciplines, including humanities, arts, social sciences, medicine, and natural sciences), and gives access to lexical patterning as well as other phraseological phenomena (Gilquin et al., Reference Gilquin, Granger and Paquot2007).

In retrospect, L2 research has recently witnessed a growing focus on chunks as potentially useful units of analysis or praxis (e.g., Ellis, Reference Ellis2003; Lewis, Reference Lewis1993, Reference Lewis2008; Schmitt, Reference Schmitt2004; Taguchi, Reference Taguchi2007), on the one hand, and the application of computers to assist language learning and teaching, on the other. Much of the research taking place in both domains, however, has been disjointed and mostly independent from each other. In other words, computer-assisted language learning (CALL) research accumulated over the years has, unfortunately, in certain cases been more of the bandwagon type and, in other cases, not adequately inspired by theoretical and experiential knowledge gained in the fields of applied linguistics or language teaching methodology. There is thus a need for research that systematically probes the interface of insights coming from the theory side (i.e., SLA and applied linguistics) as well as teaching experience and resources offered by the rapidly advancing domain of computer technology. In principle, according to Lewis (Reference Lewis1993, Reference Lewis2008), implementing the theoretical and pedagogical framework envisaged by the Lexical Approach suggests a substantial role for the use of corpora, concordance programs, and virtual learning environments. Inspired by a lexis-based view of L2 teaching, this study is, therefore, an attempt to address this lacuna and systematically explore the interface possibility of notions offered by the Lexical Approach and the affordances provided by computers. To this end, an original corpus-based computer platform was designed which enables access to various software and online resources. In addition to authentic language use and (offline) collocational options, the database has other online and software applications such as online concordancing, identifying different word meanings in different contexts of use, test-taking and feedback provision, test designing or editing, and repeated practicing of lexis through various lexical exercises.

3 Method

3.2 Materials and Instruments

After homogeneity of the groups’ general proficiency was tested, a vocabulary pretest was developed and used to measure students’ prior lexical knowledge. The target vocabulary items were from amongst those supposed to be covered later in both “lexis-based” and “grammar-based” instruction. In terms of format, the test items were similar to lexical exercises suggested or used by Lewis (Reference Lewis1993, Reference Lewis2008), for instance fill-in-the-blanks or completion, matching, short-answer, sentence recognition, and sentence-writing questions incorporating the target words. The possible maximum score on the test was 20, and the tests were checked and marked by the teacher-researcher (see Appendix). Two standardization meetings were held to agree on the range of acceptable responses in order to ensure consistent scoring outcomes. Specifically, two experts (researchers) judged and set the responses. The test was then examined for reliability and validity. The validity of the test was ensured basically through content validity. To this end, a detailed (achievement) test specification was first designed based on the content of the syllabus. Then, 65 items were constructed also benefiting from two experts’ judgments. The test was first piloted with 30 female Iranian EFL students at the same level but in a different high school. Based on the results obtained from the item analysis stage, the items with acceptable item difficulty and item discriminability were retained in the test. The intra-rater reliability of the test was estimated to be 0.92, having the rater re-examine the pretest papers after a week. The construct reliability of the pretest estimated by Cronbach’s alpha was also found to be 0.82.

To implement the Lexical Approach in the experimental classroom, the multi-purpose LexisBOARD computer platform was designed which provided students with concordances of authentic (spoken and written) language as well as other software and online affordances. The aim was to design an appropriate resource which was easy to use and localized, based on learners’ needs and interests. In Iranian high schools much emphasis is put on development of students’ EFL reading skills (Mirzaei, Rahimi & Heidari, Reference Mirzaei, Rahimi and Heidari2014). Further, as noted above, learning, practicing, and testing words in the form of Present-Practice-Produce are common activities in Iranian EFL classrooms, of course, besides grammar-based exercises. Therefore, LexisBOARD offered different software accessories and database resources whereby students could engage with L2 lexical items (needed for their educational level) within multi-word chunks in the form of an “Observe-Hypothesize-Experiment cycle” (Lewis, Reference Lewis1993: vii). Moreover, to cater for different tastes, a varied set of practice, test, quiz, and feedback affordances were available for individual experimentation with L2 lexis. Meanwhile, although a corpus does not directly tell users the meanings of lexical items or phrases, LexisBOARD was a combination of a dictionary and a corpus which provided students with different meanings of the target vocabulary based on the context of use they chose or concordance lines of words and chunks. Different word meanings and the authentic language use were extracted from the Collins COBUILD Dictionary (2006), and examples of concordance lines were taken from the BNC (2007). LexisBOARD also provided this possibility for users to be linked to the original database from which the concordance lines were taken and conduct further explorations. The rationale for designing a local lexis package was that corpus tools come in so many shapes, not all of which are necessarily compatible and useful to all groups of teachers, learners, or material designers. Thus, it was considered worthwhile creating corpus-based activities which take into account all the important variables such as age, gender, type of school, level of study, teacher’s qualifications, class size, location, nationalities, and students’ difficulties (O’Keeffe et al., Reference O’Keeffe, McCarthy and Carter2007). The figures below show different uses of LexisBOARD by these Iranian junior-high-school students; an access link to the demo version of LexisBOARD is available upon request.

In addition to different lexical meanings (see Figure 1), students also have access to different images for various co-occurrences of lexical phrases (Figure 2). Lewis (Reference Lewis2000, Reference Lewis2008) encourages the use of visual presentations such as picture or image as useful techniques in teaching vocabulary.

Fig. 1 The LexisBOARD display of lexical patterning for watch.

Fig. 2 Various images for co-occurrences of let’s.

As displayed in Figure 3, LexisBOARD provides students with examples of concordance lines for the target vocabulary. Additionally, students are given a text in which the target lexical phrases, word combinations, collocations, prepositional usage, and style are used authentically, information that is missing in Iranian EFL textbooks.

Fig. 3 Concordance lines of watch.

LexisBOARD also provides students with tests or quizzes for self-assessment. The software then checks students’ performance and provides them with appropriate feedback along with their test scores when they click on the “correct” button. Finally, LexisBOARD’s test affordance provides the test taker with the correct responses when they fail to benefit from the feedback and simply click on the “answer” button. The software also has the option to export a sample test to a word document. Figure 4 shows typical questions on different lexical phrases which co-occur with take.

Fig. 4 Sample test items on the use of co-occurrences of take.

For listening, which is one of the main techniques of the Lexical Approach, LexisBOARD plays sentences and students are asked to discriminate chunks while listening (see Figure 5). Other questions examine collocational and communicative uses of language. In these questions, students are initially assessed on the authentic use of lexical phrases, and are then asked to create a dialogue with the target phrases to notice lexical patterns and make their own generalizations about the L2 lexis.

Fig. 5 Sample listening items on discriminating chunks, and conversation-making.

As noted earlier, within the Lexical Approach, greater emphasis is put on introducing and practicing lexis, which means directing and helping students collocate words and grammaticalize from lexis (Lewis, Reference Lewis1993). In other words, teachers can exploit lexical exercises as a means for further focus on lexical phrases which determine grammar and carry more meaning than syntactical structures. LexisBOARD provides exercises for each lexical phrase, with an emphasis on language chunking and learning of formulaic expressions, and working out meanings through the context of words and collocations. LexisBOARD provides exercises to further practice the use of lexical phrases and collocations (Figure 6).

Fig. 6 A sample exercise for the practice of co-occurrences of let’s.

As displayed in Figure 7, the LexisBOARD database contains other software tools such as a survey of corpora (O’Keeffe et al., Reference O’Keeffe, McCarthy and Carter2007) which helps students access the corpus resources available electronically, options for changing visual effects such as font and color, a phonetic presentation program to improve pronunciation, and word document tools such as “save”, “delete”, and “new”.

Fig. 7 LexisBOARD software tools.