About 20 years ago, Long (2000) wrote his seminal paper on the differences between Focus on Form, Focus on Forms, and Focus on Meaning. In this article, he pointed at the lack of a widely accepted linguistic theory to form the foundation of communicative language teaching. At that time, generativism (Chomsky, 2009) was still the most commonly known and accepted theoretical linguistic framework, favoring a focus on grammar, even within a communicative approach. As a result, language instruction still predominantly focused on language as an object rather than a medium of communication. In the last decade or so, usage-based linguistic theories with its basic tenets of learning through use and exposure have found their way into the field of applied linguistics (see Tyler, & Ortega, 2018). Rather than a focus on forms, focus on form and meaning have become more and more important theoretically.

Against this backdrop, it needs to be pointed out that most foreign language (FL) teaching in secondary schools continues to be predominantly explicit and structure-based (SB) (Lightbown, & Spada, 2021, p. 166); language teaching practices in the Netherlands are no exception (West, & Verspoor, 2016). Although most FL course books used in Dutch schools claim to follow a communicative design inspired by the Communicative Language Teaching (CLT) approach and aim to develop communicative competence as advocated by different CLT proponents like Halliday (1970), Hymes (1972) and Widdowson (1978), a different picture emerges when observing classroom practices. The gradual acquisition of selected structural and lexical items constitutes the backbone of these course books, which often adopt an explicit and deductive approach to grammar, emphasize lexical and grammatical accuracy and impose the use of the L1 as the language of instruction (Popma, 1997; Hermans-Nymark, 2006; Dönszelman, 2019).

But during the last two decades, other language teaching approaches – more in line with usage-based theoretical perspectives on language learning and often labeled Dynamic Usage-Based (DUB) approaches – have been introduced in a number of Dutch secondary schools. The major tenets of a DUB approach are in line with a Focus on Form approach, but in addition, they include frequent and repeated exposure to meaningful and comprehensible input containing full chunks of language, active use of the FL, an implicit and inductive approach to grammar and a focus on productive skills in the classroom (Rousse-Malpat, 2019).

Dutch teachers are not eager to adopt such approaches. Even though they acknowledge that DUB approaches and a focus on oral production may help the learners in developing listening and speaking skills, they worry that they may do less well on reading (especially for the final exam) and writing skills (Rousse-Malpat, 2012; West, & Verspoor, 2016). Therefore, several studies have been conducted to compare the long-term effects of the more traditional SB teaching program to those of a DUB-inspired teaching program on all four language skills: listening, speaking, reading, and writing. The aim of this particular study is to compare the students’ writing skills in the two programs after six years of instruction, building on Rousse-Malpat’s work (Rousse-Malpat, 2019).

In line with generativist linguistic theory, an SB approach sees the FL as a system of structurally related elements that encode meaning. Language learning within an SB approach is inherently rule-driven (Lightbown, & Spada, 2013) and the development of FL proficiency benefits from learning to apply grammatical, morphological and phonological rules and focusing on accuracy. The structural elements of different subsystems are usually presented from simple to complex, making language structure an essential backbone of SB course books.

In line with usage-based theory, a DUB approach (see Verspoor, 2017) views language as a complex dynamic system that itself consists of different interacting sub-systems like the lexicon, syntax and morphology. The language system is complex and dynamic in the sense that it is connected to other systems in its environment, like for instance the general cognitive system and the affective system, and that its subsystems (e.g., lexicon, syntax, phonology) are not modular but interact. Furthermore, language as a complex system is dynamic in the sense that it evolves with a changing environment and changing input patterns. Second language development is the result of a dynamic interplay between internal resources like general aptitude, degree of motivation, eagerness to learn, and attention as well as external resources, including the degree of exposure or the effectiveness of an instructional approach (De Bot, & Larsen-Freeman, 2011). Language thus emerges through use and in interaction with different sub-systems that themselves foster change over time.

Key elements in language learning from a DUB perspective are repetitive exposure to meaningful input and authentic language use (Ellis, 2008; Langacker, 2000; Tomasello, 2003). When sufficiently exposed to the target language for regularities to become noticed, the learner comes to use distributional information to bootstrap knowledge, resulting in language acquisition (Onnis, 2012). Linguistic units are learned as they are “heard and used frequently and therefore entrenched, which is the result of habit formation, routinization and automatization” (Verspoor, & Schmitt, 2013, p. 354).

A fundamental difference between SB and DUB perspectives is the type of instruction it assumes to be needed for language learning. Operationalizing a SB perspective into a teaching program typically results in an explicit approach with a focus on understanding language structure and producing accurate language. Operationalizing a DUB perspective into a teaching program, on the other hand, usually involves an implicit approach to morphosyntax and an emphasis on exposure and active use of the FL and fluency in language production.

One of the language domains that has been strongly associated with SB instruction is writing. The distance between the writer and the reader in both time and space implies the absence of feedback from the reader and therefore necessitates the creation of coherent and understandable texts. FL writers generally spend a great deal of time on the three cognitive sub-processes involved in writing: planning, formulation and revision (see Hayes, & Flower, 1980). Indeed, both Fayol (1997) and Barbier (1997) have shown that the process of writing is much slower than the process of speaking, and it has been argued that especially in the area of accuracy in French FL writing, implicit knowledge might not be sufficient: “in order to ensure accuracy in the low-level aspects of the text, writers also use their explicit knowledge, especially in the case of writing in L2 French” (Gunnarson, 2012, p. 249).

Based on these opinions, it is reasonable to assume that writing skills are more likely to be affected than other skills in a paradigm shift from SB explicit approaches to DUB implicit approaches. However, this has not been supported by a number of recent longitudinal classroom studies conducted in Dutch secondary schools with free response data.

In a classroom study comparing explicit instruction and implicit instruction, Andringa et al. (2011), who present an excellent overview of the explicit-implicit debate, tested eighty-one 12–18 year old learners of L2 Dutch on their use of explicit knowledge in a free written response task and found that after four months of L2 Dutch instruction in which exposure was controlled for, there was no advantage of explicit instruction over implicit instruction on a free writing response task.

In a three-year longitudinal classroom study in the first three years of a Dutch high school (with participants aged 12–15), Rousse-Malpat (2019) conducted a large cohort study with 229 students examining the development of speaking and writing proficiency of French as a FL in several Dutch schools using the same teaching programs as in the present study: one structure-based and the other dynamic usage-based. After three years, the DUB learners outperformed the SB learners in both skills, which may have been due to the large difference in FL exposure. In one SB group, though, the teacher spoke FL French for the greater part, and in Rousse-Malpat (2019) this group was compared to a DUB group of learners with the same scholastic aptitude level on writing skills. The SB group and DUB group scored the same on holistic scores, but the DUB wrote significantly longer texts, suggesting a greater ease in language production. The DUB group also wrote longer sentences, suggesting a higher overall complexity, and used significantly more advanced morphological forms. A previous study by Rousse-Malpat and Verspoor (2012) showed that the SB groups outperformed the DUB group in terms of accuracy after one year, but this difference disappeared after two years, most probably because DUB learners need more time to discern the morphological patterns.

The classroom study by Piggott et al. (2020) involved 416 Dutch learners of English as a foreign language and investigated the effectiveness of a two-year program with explicit grammar instruction and a program without explicit grammar instruction. All 416 students used the same coursebooks, but in the implicit condition, the grammar explanations were removed, and the time left was used for more listening and reading tasks from the book. The study showed that the effectiveness of the program was associated with different aspects of language performance: while the explicit group performed better on accuracy measures in general, the implicitly taught group performed better on complexity and fluency measures. Holistic ratings showed no difference for vocabulary, but the explicit group outperformed the implicit group in relation to the grammar ratings that the experts provided.

To summarize, most CLT programs in the Netherlands are still inspired by SB views, and teachers tend to favor explicit instruction on morphosyntax and accuracy, using the L1 as the language of instruction. This reminds us of a Focus on Forms approach. Yet, several longitudinal studies in the Netherlands with free response data have shown that accuracy can also be achieved with implicit, high exposure approaches.

The current study (part of a larger study in which all four language skills are examined) explores the effect of two teaching programs for French as a FL, the same as those in Rousse-Malpat (2019), but rather than comparing the results after three years of instruction, the current study focuses on the writing results after 6 years of instruction. The research question is whether a DUB program (without explicit attention to forms) is as effective as a SB program (with a great deal of explicit attention to forms) in achieving accuracy and developing the writing skills of Dutch VWO (pre-university) students in a 6-year teaching program. Our expectation, based on the long-term studies mentioned above, is that with enough exposure to and active use of the FL, a DUB approach might work as well as an SB approach in developing writing skills, not only on overall scores but also on morphological accuracy.

The holistic scores by expert teachers and the level of writing proficiency provided by a computer program for morphosyntactic analyses of written FL French are summarized in Table 1.

Although the DUB learners scored higher on both measures, the differences were not significant and effect sizes were low.

Table 2 provides an overview of the results on CAF measures.

The DUB students produced significantly longer and more complex sentences and produced longer texts overall with medium effect sizes (Plonsky, & Oswald, 2014). For a better understanding of these results, the first paragraphs of two essays of two students will be given as an example:

These extracts clearly show both significant differences reported in Table 2. The DUB student not only outperformed the SB student with regard to the total number of words, but also to the average sentence length. Moreover, the sentences produced by the SB student were predominantly simple (no subordinate clauses) while the sentences of the DUB student were more complex (more subordinate clauses). But the extracts not only show that sentence complexity contributes to a higher perceived quality of the text but also the use of chunks. In another sub study (Gombert et al., submitted) it appeared that chunk coverage (percentage of words in chunks out of total words) is much higher in the extract of the DUB student (52 %) when compared to the extract of the SB student (35 %).

As far as accuracy is concerned, the DUB students performed slightly better than SB students when conjugating verbs, but SB students performed slightly better than DUB students when applying agreement rules pertaining to gender and number. In neither of these cases, however, the differences were significant.

In the Dutch context, teachers prefer to hold to traditional ideas about FL instruction with a great deal of explicit attention to forms because they fear that students will be less accurate if they do not. Especially for writing, the argument has been that explicit attention to morphosyntax is needed to achieve accuracy (Fayol, 1997; Barbier, 1997; Gunnarson, 2012; Ellis, & Wulff, 2015; Hulstijn, 2015). The current study was designed to test this assumption and compared learners on French writing skills in two different FL teaching programs – a structure based (SB) and a dynamic usage-based (DUB) program – after six years of high school instruction in the Netherlands.

In line with Long (2000), the SB program could be considered a Focus on Forms approach with explicit explanations in the L1 on French grammar and relatively little true exposure in the FL, especially in the first three years. The DUB program could be considered a Focus on Form approach with implicit attention to form and a great amount of FL exposure and use.

The most interesting finding was that the SB and DUB groups did not differ much at all in holistic ratings given by a group of experts, in the proficiency level score produced by Direct Profil, which included many accuracy measures, nor in the two specific accuracy measures focused on a great deal in the SB classes: subject-verb agreement and determiner-noun agreement. With regard to specific CAF measures, the results show a difference between the approaches. The DUB learners produced longer texts, which may be a general indicator for fluency, and longer sentences, which is a general indicator of sentence complexity. In a study on the same students (submitted), we also found differences in chunks. The DUB learners used longer and more lexically based chunks, and therefore relatively more words that could be classified as part of chunks than the students in the SB condition (d = 0.73, p < 0.05). This was clearly illustrated by the two examples in the previous section which showed a higher chunk coverage of the DUB student.

When we relate these findings to the literature reviewed, the results of this study clearly align with previous classroom studies conducted in Dutch secondary schools (see Andringa et al., 2011; Rousse-Malpat, 2019; Piggott et al., 2020) with regard to complexity and fluency. In all studies, the implicit teaching programs appear to be as effective as explicit teaching programs. As for accuracy in writing, Piggott et al. (2020) reported that students in the explicit condition performed better on accuracy measures, while in this study students in the implicit condition performed better on complexity and fluency measures and equally well on accuracy measures. This seems logical as the implicit group in the Piggott et al. (2020) study was tested after two years, while in this study, students were tested after six years, and as Rousse-Malpat and Verspoor (2012) showed, the implicit learners seem to take longer to internalize the more subtle morphosyntactic patterns.

Of course, the current study has its limitations. Empirical studies usually require full experimental control to allow generalization of findings, but classroom research requires high ecological validity to support teaching practice. Moreover, if this type of classroom research is widely replicated, as advocated by DeKeyser and Botano (2019), this lack of methodological rigor will be compensated as results will become more robust and more relevant for practitioners.

This study also has several strengths: First, although results were only obtained at one point in time (after six years) and development of writing skills cannot be determined over time, its duration was necessary to facilitate a comparison of two teaching programs which differ with regard to the degree of implicitness. Second, this study has a high ecological validity with participants who are not just learning for the sake of the experiment (DeKeyser, 2003), and French as a FL. As the majority of the experiments in this field so far have been conducted with English as a FL, it is important to conduct experiments with other languages as well, especially those where extramural exposure is limited. Indeed, in most existing effect studies on English, there is generally no control over informal learning at home, which makes French in the Netherlands an excellent choice for such a study as there is almost no extramural exposure normally. Despite these strengths, the findings of this explorative study should be interpreted with caution and more long-term and ecologically valid studies are needed to confirm the findings.

The current long-term classroom study suggests that general learning mechanisms like statistical learning are capable of facilitating FL acquisition through FL exposure and active use, if given sufficient time. It seems that a predominantly implicit, DUB FL teaching program might be as effective as a predominantly explicit, SB approach in terms of achieving morphological accuracy and even more effective in achieving complexity and fluency. Apparently, students do not need a great deal of explicit instruction to become accurate in writing; therefore, it might be much more useful to spend time in class on oral skills, which are seriously neglected in most FL French classes in the Netherlands (Michel et al., 2021). As another study on the same students has shown (submitted), the DUB students outperform the SB students significantly with a very large effect size (d = 2.05) on speaking skills.

Level 4 (high):

Level 3:

Level 2:

Level 1 (low):