2. Language policy

The multilingual society is a fact, both in Belgium and in the Netherlands. Minority languages like Arabic, Turkish or Hindi have a strong position, even though they are not officially supported. English and the major European languages, on the other hand, have a protected status in education. In a comprehensive investigation, Beheydt (Reference Beheydt2005) shows the discrepancy between the academic and political views on multilingualism. He states that the academic world asks for official recognition of minority languages and resists the idea that all efforts should go into majority standard language education. The official policy, on the other hand, gives absolute priority to the learning of the majority Dutch standard language, as the primary means to social mobility and integration of minority groups. Beheydt notes that, officially, multilingualism is fostered only in so far as the modern European languages are concerned. The European Commission recommends pupils learn at least two other European languages. There is only government financing of second-language education if this concerns education in the European languages. Although the Dutch government stresses that all immigrant languages be respected, state support is no longer given.

Extra & Yagmur (Reference Extra and Yagmur2004) call the language policies of European countries discriminating. Their article is a reaction to the abolishment of education in immigrant languages in 2004 in the Netherlands. In their plea for the protection of immigrant minority languages, they compare language rights to human rights. They advocate the same treatment for immigrant languages as there is for regional languages. They stress that, given the decreasing importance of nationality and country of birth in identifying multicultural groups in the European context, ethnicity and home language should replace these criteria. The large-scale ‘Multilingual Cities Project’ carried out in six large multicultural cities in Europe shows that in a city such as The Hague, 49% of primary school pupils speak a language other than Dutch at home. Turkish was shown to be the minority language with the highest vitality. Extra & Yagmur also report two case studies into minority language teaching in schools in Germany and Australia. The ‘Muttersprachlicher Unterricht’ in Nordrhein-Westfalia is state-funded and meant to stimulate multiculturalism. Languages are offered at parental request if there is sufficient demand and available teachers. Progress is measured through bilingual tests instead of monolingual German tests, and the educational inspection service supervises the education. In Victoria State, Australia, all children learn a Language Other Than English (LOTE) next to English. These two cases are presented as good examples of including immigrant minority languages in school programmes. In another publication, Extra (Reference Extra2005) emphasizes the usefulness of trilingual primary education: a language in addition to Dutch and English. Whereas the national languages of EU member states carry cultural identity, the same claim of immigrant language is rejected. In line with the recent plea of the Dutch Council of Education, Extra argues for the availability of languages other than Dutch (and English) for all pupils. In his model (2005: 16), Dutch is still the main language, with a second language of choice during four hours a week for all pupils and English in the two final years (age 10–12). He stresses the potential of local language schools in support of multilingual education.

3. The multicultural classroom

At a recent symposium, Appel & Kuiken (Reference Appel, Kuiken, Coenen and Hoefnagel2006) outline the history of the ‘young field’ of Dutch as an L2. They report on descriptive, theoretical and applied research, the forming of research groups and the rise of specific databases, such as the European Science Foundation Second Language (ESF) corpus of utterances of naturalistic, low-educated learners of Dutch. They also note the relatively few comparative studies into the effectiveness of different educational approaches and methods. As this type of research is methodologically complex and is both practically- and theoretically-oriented, funding is not easily obtained. They see a lack of empirical studies on the teaching practice itself. In the many teacher manuals and guidelines they identify a scientific mission, felt by some linguists to contribute to the solution of a societal problem: a task uncommon to linguists in other areas of language research. With interesting themes, authors of teaching methods offer the possibility to integrate the contents of non-linguistic subjects as geography or biology into language learning. This shows that there is space for content-based language learning in L2 Dutch. They also identify today's research as a new stage in the development of the L2 Dutch field: a second generation of IT materials is on the way, offering many new possibilities, as the extensive multimedia component of the method Code (Boers, Heijne, Hidma, van Kampen, Olijhoek & Van der Voort Reference Boers, Heijne, Hidma, van Kampen, Olijhoek and Van Der Voort2004) shows. However, the contribution of these new media to the speed and success of L2 Dutch acquisition is as yet unknown.

One of the central questions in the debate on education to immigrant children is the role that could or should be played by the home language in the improvement of school success in general, and Dutch language skills in particular. Turkish and Moroccan (both Berber- and Arabic-speaking) children start school with a language delay of about one half to a whole year, which often increases after primary school. In a large-scale (N = 4419) longitudinal study, Van der Slik, Driessen & De Bot (Reference Van Der Slik, Driessen and De Bot2000) investigated the relationship between the use of Dutch at home and Dutch language proficiency at school, comparing ethnic minority and non-immigrant Dutch children. Comparing Turkish and Moroccan children to other children whose home language is not standard Dutch (but, for instance, Frisian) offers a valuable insight into the role of Dutch as a home language. They found that children with a migrant background develop their skills in Dutch considerably over two years, but they lag behind the Dutch reference group. Because other groups with a home language other than standard Dutch do not show a similar trend, they conclude that the use of a language variety other than standard Dutch does not necessarily have a negative effect on Dutch language proficiency. Therefore, it is not the variety or language itself but rather the type of language used that causes the difference.

Van Steensel & Nap-Kolhoff (Reference Van Steensel and Nap-Kolhoff2005) argue that the basis for literacy is laid in the pre-school period. A quantitative analysis into the relation between pre-school playgroup participation and later school success (N = 68) showed no significant differences on literacy tests in Kindergarten between ethnic minority children who had and those who had not visited pre-school playgroups. Whereas both groups scored high on technical literacy skills, they scored far below the national average regarding comprehension literacy skills. A qualitative multiple case study of three Turkish boys in a Dutch pre-school playgroup shows that these boys were only able to engage in cognitively challenging interactions after more than a year of pre-school participation because the interactions run parallel to the children's language development. They conclude that pre-school is only effective for ethnic minority children who already have a certain command of Dutch upon arrival in these forms of pre-school education and recommend that special attention in pre-school play groups go to (immigrant) children with weak language proficiency to help them cross the threshold beyond which the interactions become more beneficial. They also stress the importance of more professional teachers who will notice when children are in the multi-word phase allowing cognitive interactions to take place.

Van Steensel (Reference Van Steensel2006) evaluates activities in a school-based program and a family-focused learning program (‘Opstap Opnieuw’) to find out if and how they contribute to language development of minority children in the first years of primary education, comparing them to language development of children not attending any pre-school activities. He found that only the second type contributed to aspects of language development, and that the effectiveness of attending a pre-school centre partly depends on two factors: organization of the centre (notably the adult–child ratio, which determines the number of cognitively stimulating interactions), and linguistic background of the child (Dutch proficiency when arriving at the centre determines their benefits of stimulating activities).

Henrichs (Reference Henrichs and Koole2006) describes the DASH-Project (Development of Academic Language in School and at Home). In this project the development of academic language in Dutch children, Turkish–Dutch children and Berber–Dutch children is the main topic of research. The differences between the home language register and the school language register, and the transition between the two is central to this longitudinal project. Through exploratory and comparative research it aims to identify the lexical and grammatical characteristics of the linguistic structure of the school language register. It is expected that the extent to which these characteristics are already present in the home context influences the ease with which young children take over the school language register. Given the exploratory approach and the all-inclusive nature of the coding scheme, a bottom–up design was necessary. Henrichs developed a coding scheme for the analysis of the linguistic structure of school and home language; the scheme will form the basis of all analyses of transcripts of interactions between mothers and children.

In a series of teacher-training studies Damhuis & De Blauw (Reference Damhuis and De Blauw2005) investigated teachers' interaction skills with (multilingual) children to find out whether classroom conversations are acquisition-oriented enough to allow for language acquisition. They consider a conversation acquisition-oriented when there is a balance between the adult's input and feedback, and the child's production. They find that current educational settings lack opportunities for production. To train teachers to allow students to actively participate in conversations, a set of interaction skills is set up called CombiList (Damhuis, De Blauw & Brandenbarg Reference Damhuis, De Blauw and Brandenbarg2004). The researchers report positive results of self-reflection through video material for trainees (teachers). A follow up study will investigate whether they have improved their interaction behaviour and created more opportunities for language acquisition in their classrooms.

Hajer (Reference Hajer2005a) also addresses teachers' interaction skills as a facilitating factor in language acquisition, especially for L2 learners of Dutch who have less opportunity for language production and feedback outside of school. She found that activating didactics, such as stimulating pupils to take part in group discussions, are used less with Dutch L2 pupils than with native Dutch pupils. She suggests that the quality of class interaction should be an item in school inspection visits. Comparing seven case studies (for details see Hajer et al. Reference Hajer, Beijer, Van Eerde, Haitsma, Riteco, Swank and Koole2006), Van Eerde et al. (Reference Van Eerde, Hajer, Riteco and Swank2006) state that teachers do not always see a role for themselves in stimulating classroom involvement. They conclude that improving teacher didactics is possible. Implicit and explicit interventions, such as video recordings and stimulated recall interviews, changed teachers' behaviour and mind set in a four-phase cycle. Teachers' linguistic awareness increases, they stimulate productive language use and eventually help develop specific aspects of students' language within the course contents. Long-term effects of the interventions or the effects for pupils are unknown. Large-scale studies are needed to measure the effect of interventions for larger groups of teachers.

Interactive teaching didactics can be applied in both language and non-language courses by means of language-focused content education, but Van Eerde (Reference Van Eerde2004) points out that most content teachers are not aware of that. Much attention is given to language in content education (see Hajer & Meestringa Reference Hajer and Meestringa2004; Van Schooten & Emmelot Reference Van Schooten and Emmelot2005; Riteco Reference Riteco2006) but this focuses mainly on L1 education.

Focusing more on the ethnic diversity of pupils in primary education, Bezemer et al. (Reference Bezemer, Kroon, de Wal Pastoor, Ryen and Wold2004) compare multicultural primary school classes in Norway and the Netherlands in terms of language teaching and learning. Drawing on observations, interviews, and school and policy documents, they investigated multilingualism and opportunities for language learning in day-to-day practices. L2 learning in the Dutch class seemed to be formal, context-independent and aimed at intentional learning by the children, i.e. explicit representations of language were transmitted. In the Norwegian classroom, on the other hand, learning was more informal and communicative, relying on incidental and context-bound language learning. These different approaches reflect the way multilingualism was dealt with. Whereas in the Dutch classroom common knowledge was presupposed by the teacher, in the Norwegian classroom it emerged from interaction between teacher and pupils. Bezemer and colleagues conclude that both ways of working have assets and problems. In the Norwegian setting, activities were often unplanned and the classroom noisy; because of this, time was not always used effectively. Explicit educational goals could enhance learning outcomes.

The predictive validity of testing children in Kindergarten for their school success was studied by Verhoeven & Vermeer (Reference Verhoeven and Vermeer2005). They report on the results of a longitudinal study on language and school development of monolingual Dutch (L1 Dutch) and bilingual ethnic minority children who were followed from entering Kindergarten at the age of 4 to the end of grade 6 in primary school at the age of 11/12. Verhoeven & Vermeer investigated to what extent an administered intelligence test, the Dutch language tests and the Turkish or Moroccan language tests predict toddlers' scores on the final primary school test, the ‘Cito Eindtoets Basisonderwijs’, several years later. They found great differences in cognitive capacity and language proficiency between L1 Dutch and immigrant pupils. Although all groups improved significantly, at the end of grade 4 (age 8 years) there were still significant differences between L1 Dutch and immigrant learners. The researchers found that school performance was strongly related to language proficiency, less strongly to cognitive capacity, and only weakly to the home situation. Dutch language proficiency yielded similar scores over the years. The researchers stress that, although the correlations between language proficiency and the final test (‘Cito Eindtoets’) are considerable, one must not have too high expectations of the predictive validity of a test at such a young age. Only 11% was correctly predicted to be prone to risk at a later age, for 15% that prognosis was incorrect, and 14% of the children showed a risk of problems that was not predicted by the language proficiency test. Although the number of erroneous predictions was more than twice as high as the number of correct predictions, this study shows possibilities for finding out whether preventing delay in school performance needs financial support, as well as for measuring educational effects, especially language development. The whole set of tests need not always be administered – the receptive vocabulary test explains 38% of variance in Cito scores in the final year. Verhoeven & Vermeer emphasize the use of a language test such as the ‘Taaltoets Alle Kinderen’ [Language test for all children] (Verhoeven & Vermeer Reference Verhoeven and Vermeer2001b) as an intervention measure for linguistically weak pupils, as well as its use as a post-test after that intervention.

4. Foreign-language education

Wilhelm (Reference Wilhelm2005) reviews the history of foreign-language teaching from 1800–1920. Until 1968, French was the more important foreign language taught at schools, but then English took over. As far as teaching methods is concerned there was a change around the 1800s. Before 1800, English was taught to small elite groups, often by a native speaker and the emphasis was on communication, but when schooling became available for the masses, the emphasis was on accuracy and systematicity in a grammar-translation approach. Not until the early 1900s did English become an exam subject, with a small English literature component.

Focusing on the role of the media in the acquisition of English, De Bot (Reference De Bot2004) reports on an international project which aims to find out which factors have a positive effect on English L2 acquisition in secondary schools in Germany, the Netherlands, Belgium and France. Among the participating Dutch schools were regular higher secondary schools and bilingual higher secondary schools. Of a total of 1570 pupils in the 15 to16 age range, proficiency, contact with English, language attitudes and family background were investigated, as well as the relation between these variables. For contact with English, the role of the media in the participants' lives was focused on. For L2 proficiency, an EFL vocabulary test was administered. To obtain attitudinal and background information, questionnaires were used. Based on the results, a model was developed showing English proficiency as being mainly dependent on attitudes towards English (which was confirmed by the results). Family background had an effect on contact with English, but contact did not have a clear effect on proficiency. When comparing nationalities, for German pupils there was a strong relation between the importance of English (attitude) and scores on the vocabulary test. For French pupils a strong relation was found between contact with English through family and the importance of English (attitude). No effect of contact with English outside of school on proficiency was found. The variables could account for few differences in L2 proficiency. De Bot recommends a more extensive test battery be used to measure L2 proficiency.

More or less parallel to the previous project, another, larger, more in-depth comparative study was commissioned by the Dutch Ministry of Education. De Bot et al. (Reference De Bot, Evers, De Quay and Van der Slik2005) compared factors that influence English proficiency of secondary school pupils in Denmark, Finland, France, the Netherlands, Norway, Spain and Sweden. More than 11,000 pupils of around 15 years of age from different types of secondary education levels were tested. A 75-item language proficiency test showed a northwest vs. south pattern in which the south scores lowest on all aspects of proficiency tested. The Dutch proficiency scores systematically reflect the three high school levels: college preparatory (vwo), higher professional preparatory (havo) and volational preparatory (vmbo). As far as teaching approaches is concerned, the Netherlands differs considerably from the other countries: frontal teaching is most prominent and 87% of pupils indicate that they never or rarely work in groups. 30% of pupils claim to have learned English through the media. De Bot et al. conclude that a lot of variation in language proficiency was found among the eight investigated countries, but differences were largely explained by the global factor ‘country’. Despite the extensive set of variables investigated, the specific aspects in which countries distinguish themselves did not emerge. The role of the media in countries seems like a plausible explanation, but the data do not support this. The fact that ‘country’ as a factor explains so much more than all the other chosen variables chosen indicates that there are other factors behind that global factor which have not been considered. Interestingly, the factor ‘English in education’ did not have a significant influence on language proficiency, apparently because of the lack of variation in teaching methods, especially in the Netherlands and less so in the Scandinavian countries and Spain.

For people in the Netherlands, learning English has priority, but in Belgium, where multilingualism is considered an important asset, French is the important L2 for speakers of Dutch. In Brussels, knowledge of French and Dutch (Belgium's two main national languages) and English is a pre-requisite for good career opportunities. In this context, students' attitudes towards languages and language learning are of interest to both language teachers and educational policy makers, since these attitudes are assumed to have an (indirect) influence on the success rate of second and foreign-language acquisition. Drawing on the outcomes of an earlier study by Lochtman, Lutjeharms & Kermarrec (Reference Lochtman, Lutjeharms, Kermarrec, Housen, Pierrard and Van de Craen2004), Lochtman & Ceuleers (Reference Lochtman and Ceuleers2005) investigated the relationship between language perception, attitudes and language learning motivation in a multilingual environment, assuming a learner's perception of a target language affects attitude. In the first study (Lochtman et al. Reference Lochtman, Lutjeharms, Kermarrec, Housen, Pierrard and Van de Craen2004), the attitudes and motivations towards the five languages taught in Belgium (Dutch, French, English, German and Spanish) of 191 French and Dutch economics students from Brussels were measured with a questionnaire. Lochtman et al. found that language perception reflects stereotypes and has nothing to do with learning attitudes. They also found that Dutch and French students did not differ significantly in their opinions of English, German and Spanish; they did differ significantly, however, in their attitudes towards their L2 (Dutch/French): French learners were less positive. In the second study (Lochtman & Ceuleers Reference Lochtman and Ceuleers2005), 235 secondary school pupils from Brussels from both French and Dutch schools ticked five-point scales pertaining to their attitudes to foreign-language learning and answered the question how much effort they wanted to put into learning a foreign language, i.e. how motivated they were. Despite the French pupils' more negative attitudes towards Dutch, monolingual Dutch and monolingual French pupils were equally motivated to learn the L2. Finally, the researchers found that contact with the target language has a positive influence on how learners rate their own proficiency, and thus on linguistic confidence. For language teachers and policy makers, this study indicates the importance of taking different learner backgrounds into account, as well as the importance of language contact for self confidence and a positive learning attitude.

Over the last decade or so, bilingual education has become popular in the Netherlands at the elementary, secondary and higher education level. Even though much research abroad shows that early foreign-language education has no detrimental/negative effect on L1 development (Herder & De Bot Reference Herder and De Bot2006), developments in bilingual education have caused some political and societal debate. For example, in a newspaper column, Appel (Reference Appel2003) dismisses the concept of early English as another educational innovation for which there are no qualified teachers and which will harm especially the significant numbers of pupils who are already struggling with severe Dutch language delays. The findings in the studies reported on below seem to confirm Appel's worry about qualified teachers; however, there is no evidence that minority students are negatively affected by bilingual education.

Supported by the Rotterdam Early Bird project, an ambitious bilingual program for primary school children with 3 hours a week of English medium teaching, Goorhuis-Brouwer & De Bot (Reference Goorhuis-Brouwer and De Bot2005) measured the language proficiency of 88 pupils from two schools. Based on Reynell test scores for language comprehension (administered in the first and second school year), they found no negative effects of early foreign-language education on L1 proficiency. Pupils with a non-Dutch background even develop their Dutch skills almost beyond their initial delay. In a less ambitious bilingual program, Aarts & Ronde (Reference Aarts and Ronde2006) found in their classroom observation study that even though the teaching is playful, with activities according to themes and the help of audiovisual materials, pupils mainly listen (60% of teacher behaviour consisted of frontal instruction and asking for attention) and there is not much room for interaction or production by pupils. At the end of class two, children could understand school instructions and assignments in English, but could not form sentences. Like Goorhuis-Brouwer & De Bot, they found no significant effect of home language on the children's language development. The researchers suggest that more attention be paid to the pupils' language production. Finally, the quality of English used by teachers can be improved, and pupils' development ought to be monitored.

Most bilingual or early L2 programs in the Netherlands are concerned with English as the L2, but especially along the German border there are also some schools that have experimented with L2 German. Admiraal & Jörg (Reference Admiraal and Jörg2002) report on the results of the Dutch primary school pupils who took part in an international project called EUROBABEL. The project had both cultural and linguistic goals. Four primary schools, two Dutch and two German, took part in the project, which lasted three years (1995/96, 1996/7, 1997/8). Each Dutch school was given a German partner school and teachers were exchanged so that native speakers were teaching in the project. At the start of the project, the Dutch children were about 4–5 years old; the German children were about 8–9 years old. German was introduced only in the non-linguistic subjects. Because of the limited number of pupils who participated and incomplete data available, only tentative conclusions could be drawn, but they found no negative effect on pupils' L1 language proficiency or calculation skills.

Despite the fact that bilingual education has become more popular at some schools and English became a compulsory subject in the final two years of primary education in 1986, Oostdam & Van Toorenburg (Reference Oostdam and Van Toorenburg2002) show a rather dismal view of English as a compulsory subject in primary education in the Netherlands as a whole. When in 1991 and 1996 pupil performance was measured, performance was lower than expected: fewer than 50% of pupils scored ‘sufficient’. In 1993 and 1998 learning targets were formulated, but as these were only general guidelines, English teaching both in primary schools and at teacher training institutes has had but a sober existence. To investigate these issues further, the authors administered questionnaires among 147 teachers in primary and (the first year of) secondary education, and 295 pupils in the first year of secondary education. They conclude that pupils enter high school with very different levels of proficiency, which high school teachers usually ignore in their teaching, frustrating the more proficient students. They argue that the lack of professionalization among primary school teachers, primary schools and teacher training programs is due to the fact that there are no official standards to be met at the end of primary school and primary schools are not held accountable for the L2 proficiency of their students.

Whereas bilingual teaching at primary school has not yet really taken off in the Netherlands, the number of secondary schools offering bilingual programs is growing rapidly, starting with one school in 1987, 15 in 1997, and 100 in 2007. In addition to Content and Language Integrated Learning, these schools provide an international, especially European, perspective in school subjects. Roughly half of the curriculum is taught in English from the start. Programmes start in secondary school grade 1 (12-year-olds) and last four to six years. The subjects taught in English vary from school to school. English, history, geography, biology, computer studies, art & design and physical education are almost always among them. At the moment, bilingual education is mainly restricted to pre-university education. Both native and non-native speakers of English teach the subjects, and authentic English materials are used. Studies so far indicate that bilingual education leads to better linguistic skills without detrimental effects on content learning (Admiraal, Westhoff & De Bot Reference Admiraal, Westhoff and De Bot2006). De Bot & Maljers (Reference De Bot and Maljers2005) attribute the success of bilingual secondary education to a carefully worked out quality control system (which includes a set of standards and regular school visits by an expert team) developed by the network of bilingual schools themselves with the help of the European Platform (<http://www.europeesplatform.nl> accessed 07/12/2007).

One of the winners for the European label of innovative language teaching is the project Slash 21. In her article, Mulder (Reference Mulder2003) describes the workings of this new system, which focuses on foreign-language education. The program consists of three curricula: science, humanities and foreign languages, which contain all regular school subjects. The foreign-language curriculum is two-fold. It starts with an initial, intensive training period of 12 weeks that lays the foundations for the language: for English and German in the first year, and for French in the second year. The intensive part of the curriculum is followed by a maintenance program that continues language education and follows the themes of the science and humanities curricula. Interestingly, the first results show that in the initial 12 weeks the core targets as set for the first two years of secondary school are achieved. In this innovative program, the target language is the language of communication, although this is built up gradually for French. Mulder states that the developed materials will be adjusted after several test rounds and then made available to other interested users by the Stichting Leerplan Ontwikkeling (SLO).

Janssens et al. (Reference Janssens, Van Mensel, Housen and Pierrard2005) studied the development of productive skills in L2 Dutch. They considered the case of French-speaking pupils in Dutch-language education in Brussels. In the bilingual Belgian capital Brussels, there are two independent education systems: a Dutch-language education system and a French-language education system. Pupils and parents can choose between both systems freely and independently of their home language background. Many Francophone parents prefer to enrol their children in Dutch-language schools in Brussels, because they believe this will stimulate their bilingual development. As a consequence of the massive inflow of French-speaking pupils in Dutch-language education in Brussels (52%), an atypical form of education has emerged which they call ‘immersion-like education’. In this paper Janssens et al. discuss a longitudinal study which looked at the development of L2 Dutch productive skills of native speakers of French and native speakers of Dutch across three consecutive school years. They traced the development of accuracy, complexity and fluency by means of repeated measures analyses of variance. They found that language productivity and complexity of L1 Dutch and L2 Dutch pupils develop evenly over time, but L2 Dutch pupils have less accurate L2 Dutch oral skills. However, L2 Dutch pupils reach the same level as their L1 Dutch classmates towards the end of their school career. No differences were found for complexity and fluency of Dutch language use; nevertheless, some differences in accuracy persist. They conclude that this form of unstructured immersion-like education, where no adaptations for L2 Dutch learners are made, successfully stimulates the development of bilingualism. It has to be added that the social and societal context of the acquisition process of these L2 Dutch learners is wholly different from that of immigrant L2 Dutch learners.

5. Reading and writing

In a comprehensive meta-analysis of writing classes supported by computer practice, Van Schooten, Fukkink & De Glopper (Reference Van Schooten, Fukkink and De Glopper2004) focus on the role of computer programs as a means of instruction supporting writing education. Such programs offer exercises for specific skills, analyse writing products or stimulate dialogical aspects of the writing process. To find out whether such programs have proven useful for the teaching of writing, Van Schooten et al. compared empirical studies into the effect of computer supported writing programs. They found a moderate effect, which is interesting as students had worked with the programs for only a little over 21 weeks. The size of the effect ranged widely. Writing focused on structure profited greatly from computer supported learning (CSL); narrative writing profited least, perhaps because feedback on creative elements in writing was outside the reach of the computer system. Furthermore, programs aimed at the writing itself caused smaller effects than programs aimed at the preparatory stages of writing or at revision. However, because some variation in effects could not be explained fully, they suggest more research is needed to understand the effects of CSL.

Van Gelderen, Schoonen & Stevenson (Reference Van Gelderen, Schoonen and Stevenson2003) report on a research project in which several universities and research institutes take part: NELSON (acronym for ‘Nederlanders en Engels, Lezen en Schrijven in het Onderwijs’: Dutch and English, Reading and Writing in Education) has as its topic the development of L1 and L2 reading and writing comprehension in secondary education. The project, funded by the Dutch Organization for Scientific Research (NWO), started in 1998 and ran until after 2005. It aimed at integrating Dutch and English school curricula, a goal for which teaching materials for the training of reading and writing skills are developed. The project was linked to current developments in Dutch secondary education, such as ICT applications in language education, differences between students with Dutch L1 and L2, and individual differences in learning strategies. Since 1999, about 400 students from eight schools were involved in the project. The intervention studies concentrate on improving reading and writing through computer training in speeded lexical access and speeded lexical retrieval. First, a baseline performance per pupil is obtained through a separate programme, followed by training sessions. Afterwards, the pupils are measured again with the separate program to check for progress.

Van Gelderen et al. (Reference Van Gelderen, Schoonen and Stevenson2003) report on the first two studies within the NELSON project. The first study investigated the development of reading and writing skills in L1 and L2 for 3000 pupils, and focuses on transfer of L1 strategies and on the influence of (automatized) linguistic knowledge. It consists of a reading comprehension test; writing tasks; a meta-cognitive knowledge test on reading and writing strategies; vocabulary, spelling and grammar tests; and, finally, four types of reaction time tests per language. A multiple regression analysis showed that reaction times had no additional value in explaining reading and writing skills; they depended mainly on meta-cognitive and linguistic knowledge. The second study investigated cognitive reading and writing processes for 24 pupils. Think-aloud protocols were used during reading and (computer) writing tasks, and the design included on-line word recognition tests. The researchers report on results from the reading tasks. Results show that in L2 reading, logically, more problem-solving strategies are used than in L1 reading, and that word recognition is related to use of linguistic strategies. Attention spent on L2 reading strategies did not negatively influence attention spent on content. The important effect of meta-cognitive skills of L1 reading and writing skills on L2 skills supports the transfer hypothesis. These outcomes indicate foreign-language knowledge as a condition for transfer. No clear support for the automatization hypothesis has been found.

The use of training word recognition (i.e. the automatization of vocabulary knowledge) for reading and writing is investigated in studies 3 and 4 of the NELSON project. In addition to the more descriptive first and second studies, the third and fourth studies develop training programmes (see Fukkink Reference Fukkink2003 above). Snellings, Van Gelderen & De Glopper (Reference Snellings, Van Gelderen and De Glopper2003) report on the fourth study within the NELSON project. To trace effects of faster word recognition on L2 writing products, pupils were trained in an intervention experiment in word recognition in English. The researchers found that increased word recognition speed led to a greater use of the trained words, and the content aimed for was better formulated in one of the two experimental groups. However, no improvements in overall text quality were found as a result of faster word recognition. Thus, on the basis of this study, no conclusions regarding the importance of word recognition speed for the writing process can be made.

Two papers deal with the Educational project ‘Concourse’, set in the collaborative IT environment SURF (http://www.surf.nl, accessed 07/12/2007). Stassen, Baudoin & De Jong (Reference Stassen, Baudoin, De Jong, Daems, Rymenans and Rogiest2002) and Van den Berk, Klein Gunnewiek & De Graaff (Reference Van Den Berk, Klein Gunnewiek and De Graaff2002) report on the development and implementation of this environment aimed at the improvement of foreign-language writing education. One of the sub-projects, ‘World Wide Writing’ won a European distinction in 2002 for Most Innovative Foreign Language Project of 2002. The collaborative learning model solves the problem of time-consuming teacher feedback by having students write peer assessments. In an on-line writing centre, students can train in academic and business writing skills in five languages: Dutch, English, Spanish, French, and German. During the writing process they receive feedback from an editorial board consisting of professional colleagues and students both at home and abroad. Van den Berk et al. set out to evaluate the new IT-based learning method ‘De Werkplaats’, developed by the German department of the university of Utrecht as part of the SURF Educational project ‘Concourse’. ‘De Werkplaats’ differs from other digital learning environments because it integrates writing assignments for language acquisition courses and content-based courses, and because it is primarily a place for collaborative learning (and writing).

The evaluation shows that (oral and) written language skills of students improve. Students also show considerable progress in the formulation of research questions, writing in different genres, giving feedback, and planning work. Moreover, students are positive about the method and its effect on their own learning; they especially mention the value of feedback. Students work more intensively and in a more organized way, but do not always experience collaboration as an effective method. They claim that coaching is needed for an effective balance between individual work and group work. While Van den Berk et al. (Reference Van Den Berk, Klein Gunnewiek and De Graaff2002) define ‘De Werkplaats’ as successful, with its positive effect on writing skills and the appreciation of students, there is room for improvement in the didactic approach of coaches.

Blok-Boas & Materassi (Reference Blok-Boas and Materassi2004) report on a successful experiment that promoted activating didactics, such as the stimulating of class participation, in an Italian L2 writing course. The course consisted of several task-focused activities leading to a final writing product and its presentation, at level B2 of the Common European Framework of Reference (CEFR). The course uses the Internet, video recordings and blackboard. Outcomes are promising: the recursive (group) discussions and peer reviewing have led to a critical attitude among students, more commitment and insights into students' own skills, and more commitment (as demonstrated by adherence to deadlines). Because all writing of texts and commentaries takes place at home, class hours are used more profitably. The course structure is less suitable for groups of more than sixteen students because of the time spent on presentations and watching video recordings.

Kuiken & Vedder (Reference Kuiken and Vedder2004) look at the relationship between cognitive task complexity and linguistic performance in writing. They test Robinson's Cognition Hypothesis of task-based L2 development that cognitively demanding tasks lead to the use of lexically and syntactically more complex language (Robinson Reference Robinson and Robinson2001a, Reference Robinsonb; Reference Robinson2005). Kuiken & Vedder report on an experiment carried out among 51 Dutch university students of Italian as a second language. In two writing tasks, cognitive task complexity was manipulated by varying the number of elements to be described and the required reasoning demands. For L2, no effect of task complexity on the syntactic and lexical complexity of the writing product (text) was found; however, fewer mistakes were made in the more complex task. For L1, in the non-complex task, a higher syntactic complexity was found than in the complex task. The researchers contribute the absence of an effect of task complexity on L2 linguistic quality to the low L2 proficiency level.

Looking at L1 Dutch, L1 English and L2 English writing, the study by Smits (Reference Smits2002) is based on a contrastive corpus analysis and the results of a psycholinguistic experiment. It proposes a functional typology for complex beginnings, investigates the effectiveness of specific types of such beginnings in particular contexts, hypothesizes reasons for differences between native and learner complex beginnings, and investigates how these are produced and to what extent their form and use are influenced by production processes. Smits found that L2 users have more difficulties in using and recognizing complex beginnings than L1 users. Learners of English most often use rhetorical ‘stepwise orientations’ (Smits Reference Smits2002: 244) whereas L1 users also used other kinds of complex beginnings such as ‘complex orientations’ (ibid: 244). In line with findings by Kuiken & Vedder (Reference Kuiken and Vedder2004), Smits concludes that the lower language competency of L2 learners may obstruct adequate production of complex beginnings, irrespective of their native language writing skills. As this study leaves many questions unanswered, further research is needed. In this vein, a Ph.D. project by Van Beuningen on the effect of experimental writing tasks on writing products has recently started at the University of Amsterdam.

Reviewing recent studies, Hulsker (Reference Hulsker2003) identifies seven factors relevant to (English) reading proficiency such as word knowledge, grammatical knowledge, the understanding of textual structure and of implicit information. She integrates those as subtests into an experimental diagnostic test of English reading proficiency. 449 Pupils took both the diagnostic test and a general reading proficiency test. An Item Response Theory analysis showed that vocabulary knowledge and the understanding of sentence structures were related strongest to general reading proficiency scores, and she suggested that grammatical knowledge unjustly receives less attention than vocabulary knowledge in reading proficiency research. No significant relations were found between reading proficiency, background knowledge, or monitoring ability, factors that need not be included in a diagnostic reading proficiency test, according to Hulsker. A factor analysis showed that each subtest measures a common aspect of reading proficiency as well as something independent of what the other subtests measure. No clusters of factors were found: individual pupils have problems with different combinations of factors. This means that no problem profiles can be drawn up; training ought to focus on individual needs. Hulsker recommends training the understanding of textual structure (e.g. knowledge of cohesion elements). The finding that the understanding of explicit information correlates stronger to reading proficiency than the understanding of implicit information supports this recommendation.

Hacquebord (Reference Hacquebord2006) discusses the results of a diagnostic language test among 2718 pupils from 17 schools of different levels (from ‘vmbo’ up to ‘vwo’). She focuses on vocabulary knowledge and reading comprehension of L1 Dutch vs. non-L1 Dutch and dyslexic vs. normally-developing first-year pupils (12/13 years old). An electronic test of text comprehension and a vocabulary test for the beginning of secondary education were used (published at http://www.diataal.nl as Diatekst and Diawoord). The vocabulary test showed that about 14% of students in every school type score between 1 and 2 standard deviations below the minimum norm for that type. For 80% of students, vocabulary knowledge is related to text comprehension. Interesting results come from the 20% of students for whom no relation between scores was found. Half of these students score satisfactorily on vocabulary knowledge but unsatisfactorily on text comprehension. They are diagnosed as having a reading problem and need reading training. The other half scores satisfactorily on text comprehension but insufficiently on word knowledge; they have a language problem and need training in vocabulary, e.g. how to deduce word meanings from texts. Hacquebord shows that non-L1 Dutch students in all school types have a vocabulary delay. Their text comprehension problems will mainly be language problems. Dyslexic students have no vocabulary problems; they mainly have a reading problem. What is important in the light of language testing research is the development of a student track system to monitor students' linguistic development, a system which is currently being developed by Diataal.

6. Word knowledge and vocabulary

The role of deep word knowledge in academic performance is the starting point of a study by Verhallen & Van der Zalm (Reference Verhallen and Van Der Zalm2005). They report on an empirical study into the effects of a new vocabulary teaching approach in Kindergarten. The word learning results of two methods were compared: one not specifically aimed at language or vocabulary and the recent vocabulary learning method ‘Met Woorden in de Weer’ (Working with Words) (Nulft & Verhallen Reference Nulft and Verhallen2002), aimed at broad and deep word knowledge. 94 children took a vocabulary size test, and a categorization test evaluating deep word knowledge. Children trained with ‘Met Woorden in de Weer’ improved both in breadth and in depth of vocabulary. However, the categorization test consisted of only five items and its reliability was low. New research using other, more extensive deep word knowledge measures (for example, a definition task or word association format) is needed. Future work should also test whether training such as the ‘Met Woorden in de Weer’ instruction increases general deep knowledge or just the specific meaning relations from the themes taught.

The importance of investigating different aspects of (deep) word knowledge is shown by Van de Rhoer & Vermeer (Reference Van de Rhoer and Vermeer2005). They report on an MA dissertation (Van de Rhoer Reference Van de Rhoer2004), which tested the assumption that explicit teaching of hierarchical relations in vocabulary input has a positive influence on network construction in the lexicon and the retention of vocabulary. In a quasi-experimental design, they investigated whether a group of pupils receiving instruction and exercises explicitly aimed at thematic, associative and hierarchical relations demonstrated a better retention of words than a group of pupils who received instruction and exercises focusing only on thematic and associative relations. Through a pre- and post-test design, they found a significant improvement for both groups of pupils, but no significant difference between the groups. They state that hierarchical, thematic and associative relations are all equally important for word retention and that extra focus on hierarchical relations is not necessary.

Haest & Vermeer (Reference Haest and Vermeer2005) challenge the assumption that depth of word knowledge is an important determiner of reading comprehension. They investigated various aspects of word knowledge and their relation to text comprehension skills (testing 197 monolingual and bilingual 10-year-old children). Bilingual children performed significantly lower on all tasks; the greatest differences were found in breadth of vocabulary, the lowest on the depth task. In addition, the breadth task showed the highest correlations with text comprehension, the depth task the lowest. A (stepwise) multiple regression analysis revealed that whereas the breadth task explained 38% of the variance of the text comprehension scores, the depth task only 2%. Haest & Vermeer conclude that text comprehension is mainly determined by vocabulary size.

Inspired by the psychometric tradition of intellectual assessment, traditional norm-referenced language tests rely heavily on children's ‘world knowledge’. Gerrits (Reference Gerrits2005) posits that when test takers differ in their exposure to concepts, words and activities, as is often the case with children from different ethnic, cultural, or economic backgrounds, any assessment tool that taps the child's existing store of knowledge runs the risk of confusing ‘difference’ with ‘disorder’. In her article on the testing of language delay and language impairment, Gerrits discusses problems and recent solutions for the use of norm-referenced testing, with a focus on processing dependent procedures such as non-word repetition.

Drawing on their involvement in developing curriculum materials for L2 Dutch, Van de Guchte & Vermeer (Reference Van de Guchte and Vermeer2003) discuss the importance of word selection for curriculum materials concerning vocabulary, which is often a problem for both teachers and material designers. Using outcomes of earlier work (Verhoeven & Vermeer Reference Verhoeven and Vermeer2001a) in which the ‘Taaltoets Alle Kinderen’ [language test for all children] (Verhoeven & Vermeer Reference Verhoeven and Vermeer2001b) was developed, they demonstrate the relevance of careful vocabulary selection. They discuss criteria for vocabulary selection, and a measure for word selection is put forward: a geometrical mean, which reflects a word's range as related to word frequency, and corresponds to learners' age and level. They emphasize that word lists remain only an indication of suitability and that determining which words are taught also depends on the teaching situation and medium (books, computer or interaction).

Vermeer (Reference Vermeer, Coenen and Hoefnagel2006) argues for structured, knowledge-based education rather than what he calls the ‘fashion’ of student-focused or process-based education. He states that, for the process-based approach to be successful, learners need to have a basic knowledge level. In particular, bilingual children who receive little or no Dutch language input at home will profit more from a structure-based approach than from a developmental approach. They need the knowledge and teaching material available at school, rather than a coach who stimulates and elaborates the linguistic knowledge and skills acquired at home, as is the case for L1 Dutch children. Vermeer reviews two studies from primary and secondary education that compare the effects of implicit and explicit vocabulary teaching on monolingual L1 Dutch and bilingual L2 Dutch students, which show that significantly more progress was made through the explicit teaching of vocabulary.

Working from a constructivist approach to language learning, which holds that language is learned through simple learning mechanisms rather than governed by innate language universals, Blom (Reference Blom2003) presents a survey of corpus-based case studies into pre-selected types of lexical patterning like idiom and collocations. Blom argues that to do justice to the phenomenon of conventionalism in language use, the teaching methodology should be essentially text-based, but rather than requiring students to memorize the texts, they should be instructed ‘to study them thoroughly by reading and listening, so as to make the students experience the urgency to meet communicative demands’ (p. 32). She shows that the ‘Delft Method’ (Montens & Sciarone Reference Montens and Sciarone1994) for L2 learners of Dutch, which provides a thorough study of texts, constitutes a viable method to acquire proficiency in the target language.

The well-documented fact that L2 readers have less automated word recognition skills and more difficulty with linking letters to sounds led De Milliano, Droop & Damhuis (Reference De Milliano, Droop, Damhuis and Koole2006) to investigate the effect of multimedia in support of word recognition and reading. They looked at the effect of reading supported by speech on the text comprehension of young readers in digital learning environments as opposed to comprehension through reading without speech. Participants had different reading proficiencies and spoke Dutch either as an L1 or L2. The researchers found that, overall, speech-supported reading did not lead to better, nor worse, text comprehension, for L1 or L2 readers nor weak or strong readers. The limited vocabulary of young readers may have a greater influence on reading comprehension than the mode of text input. The researchers add that the addition of audio material to text reading contributed positively to pupils' motivation, although this might well be a test artefact.

Mondria & Wiersma (Reference Mondria and Wiersma2004) investigate the commonly held belief among foreign-language teachers that it is best to learn words from a bilingual list in two ways, from L1 to L2 (productive) and L2 to L1 (receptive). However, as the Dutch final exams of secondary school (‘centraal eindexamen’) consist of a reading comprehension test that requires mainly receptive knowledge, they wondered if learning both ways is worth the effort. They administered a receptive and a productive retention test to 198 fourteen-year-olds both immediately after they had learned 16 new French words and two weeks later. Looking at long-term retention effects (the second test), they found the combination hypothesis not confirmed. They conclude that learning both receptively and productively is only useful when the target is both receptive and productive knowledge.

In another look at learning from word lists, Mos (Reference Mos2004) investigated the influence of clustering on vocabulary learning. Central to her study is the distinctiveness hypothesis, which posits that words are best learned together if they are distinct; otherwise, cross-association may occur, with the result that words are confused. She reviews studies showing that learners have greater difficulty with learning semantically-related words from the same category than with learning unrelated words. In her experiment, L2 learners of Spanish were offered computer training in which they learned three clusters of eight Spanish verbs: semantically related verbs, orthographically-related verbs and unrelated verbs. The semantically-related words turned out most difficult to learn (i.e., most learning trials were needed) as was predicted, but no difference was found between the orthographically-related and the unrelated words. Further study needs to include the effect of thematic word clustering on vocabulary learning.

Hulstijn (Reference Hulstijn and Bossers2004) paves the way for a somewhat forgotten aspect of foreign-language learning and teaching: language reception skills. He argues that training word-by-word reception is a form of implicit learning that leads to the automatic acquisition of sounds, words and word groups, which is a condition for spoken language comprehension. Given that language comprehension is only possible if a listener has perceived what has been said, a learner has to be able to recognize words in fluent speech. He proposes five steps in the training of word-by-word perception: listening for comprehension (without reading support), familiarizing yourself with unknown words, accurately perceiving fragments, first without and then with a transcript, and repetition of the previous steps for automated perception. He argues that the practice of fluent, accurate perception is useful for both beginner and advanced learners. For educational authors and teachers he advises the use of both new texts for learning new words and of texts with familiar words for the training of reception skills for known words.

Focusing on second-language listening comprehension, Poelmans (Reference Poelmans2003a, Reference Poelmansb) distinguishes between hearing, word recognition and comprehension. She investigates the effect of training lower-order skills against training higher-order skills by means of an aural lexical decision task. She measured the relation between L2 Dutch listening comprehension skills and vocabulary knowledge or degree of automatic word recognition (WR). No straightforward relation between vocabulary knowledge and WR, nor between vocabulary and listening comprehension was found in on-line and offline tasks performed by L1 Dutch and L2 Dutch learners. The effects of two different training sessions on listening skills were investigated. No training effects were found: varying listening success in both learner groups resulted. Speed and correctness of WR was found to be an accurate measure of automatization of WR. Poelmans concludes that there exists a continuum rather than a dichotomy between automated and controlled (word recognition) processes.

To what extent can WR in a foreign language be automatized and does that help foreign-language reading? A central assumption here is that WR is a condition for effective reading. Fukkink (Reference Fukkink2003) describes the effects of an experimental lesson series on learners' WR (lexical access), reading speed and reading comprehension. The training sessions used words from English texts and consisted of computerized exercises: a cloze test, two match tests and a translation test, allowing for an active and flexible training of semantic and grammatical word aspects. The training did indeed lead to faster WR, but not to a qualitative improvement of WR. Fukkink found that reading speed and text comprehension did not improve and concludes that fluency of WR does not guarantee success in reading comprehension. However, he does recommend stimulating automatization of WR in the first stage of foreign-language learning; in subsequent stages it may be integrated into the vocabulary curriculum. However, this study cannot justify too great a focus on specific automatization training in foreign-language reading education.

Idiomatic expressions make up an important part of the vocabulary of every language and should therefore be taught in foreign-language education. In their paper, Eyckmans, Boers & Beeckmans (Reference Eyckmans, Boers and Beeckmans2002) show that idiomatic expressions can be learned effectively, using the insight from cognitive semantics that a large part of figurative language use is motivated rather than arbitrary. To learn idiomatic expressions, students should achieve a heightened metaphoric consciousness or language awareness through insight into the literal or original meaning of the idiomatic expression, which then makes the expression concrete and facilitates memorization. The results of their experiment suggest that mental imagery can be a powerful mnemonic strategy, and secondly, that this strategy generates superior recall, especially (though not exclusively) with regard to etymologically somewhat transparent figurative idioms, even though processing these may require relatively little cognitive effort.

The importance of the acquisition of multiword combinations for oral language skills was the concern of a study by Eyckmans et al. (Reference Eyckmans, Boers, Stengers, Bauwens and Koole2006), who compared the assessment of speaking skills of two groups of Dutch students. The experimental group, who followed a phrasal teaching method based on a lexical approach, scored significantly better than a control group in fluency and lexical range. The findings show that teaching approaches using multiword combinations lead to a higher score in speaking skills. Recognizing the importance of strategical competence for interpreting, this study has led to the initiation of a research project on the value of phrasal learning for (the training of) interpreters.

The relatively recent upsurge in the study of vocabulary in foreign-language learning research has gone hand in hand with a growing interest in testing lexical development (see, for example, the WAT test by Schoonen & Verhallen Reference Schoonen and Verhallen1998). Meara's ‘Yes/No Vocabulary Test’ (Meara & Buxton Reference Meara and Buxton1987) is the most authoritative vocabulary size test and used as a standard instrument for research and assessment purposes (Eyckmans Reference Eyckmans2000). Eyckmans (Reference Eyckmans2004) evaluated the ‘Yes/No Vocabulary Test’ as a measure for the receptive vocabulary knowledge of French-speaking learners of Dutch. She states that the test's validity is threatened by the response bias that the test format seems to provoke in the test takers: the high false alarm rates that she finds make the test unreliable. A number of variables (the test instruction, the computer interface, the test content) that could contribute to the validity of the ‘Yes/No Vocabulary Test’ are investigated on the basis of theoretical considerations as well as experimental data. Eyckmans presents an alternative vocabulary test (the ‘Recognition Based Vocabulary Test’) that retains many of the characteristics of the ‘Yes/No Vocabulary Test’ but outperforms it in circumventing the response bias problem. Eyckmans concludes that validation research should include the interaction of the test task with the characteristics of the test takers and the testing context.

7. Grammar and syntax

Van Boxtel (Reference Van Boxtel2005) focuses on the learning of grammatical constructions by late language learners, investigating the relationship between a native-like level of proficiency on the one hand and the typological distance between the language pairs involved (Dutch–German, Dutch–French and Dutch–Turkish) and participants' background characteristics on the other hand. She presents a study in which an assessment was made of the (implicit) L2 Dutch grammatical knowledge of 43 native speakers of German, French and Turkish, who came to the Netherlands after the age of twelve. Using very strict criteria to specify the native speaker range, she found that in each L1 group a number of learners had reached native-speaker level in L2 grammar after the age of twelve. Factors such as input, attending Dutch classes and age of arrival had limited influence. However, there did seem to be a meaningful relation with level of education, proficiency in some other language and pleasure in learning languages. Moreover, many participants had a background in linguistics (for example, they were language teachers). For second-language teaching the indication was that raising language awareness to stimulate students' noticing structures is very important for morphosyntactic aspects of learning.

Van de Craats' (Reference Van de Craats2005) investigation into the acquisition of finiteness leads to insights into the order of acquisition in L2 Dutch. The acquisition of finiteness, more particularly the morphological marking of finiteness, is a well-known obstacle in L2 acquisition by adult learners. In her paper, Van de Craats deals with an interlanguage stage in the acquisition of Dutch by Turkish and Moroccan learners producing a noteworthy pattern that occurs neither in the target language, nor in the learners' native language. Informants were recorded three times over five months for several hours during which they carried out the productive task of retelling a wordless animation film. The learners use lexical markers (free morphemes) as precursors of fully inflected thematic verbs (as in hij is liegt ‘he is lies’). In this pattern, the lexical marker ‘is’ (copula verb, third person singular) is linked either to an infinitival, an inflected form, or to a second, identical form. Van de Craats argues that this lexical marker is deprived of meaning and that its function is primarily syntactic. The inference is that syntax precedes morphology, especially free morphology, in adult L2 acquisition.

To what extent L1 interference causes stagnation in L2 acquisition was the question central to a study by Van de Craats (Reference Van de Craats, Coenen and Hoefnagel2006). Her informants were low-educated adult L2 learners who learn Dutch in an educational setting (an L2 Dutch course). Van de Craats predicts that for these L2 learners, just as for unguided learners, the characteristics of the mother tongue are the starting point of the L2 acquisition process and a possible factor in stagnation. A longitudinal experiment was set up in which Turkish and Moroccan women carried out receptive and productive tasks in Dutch over a period of 14–16 months. Both free and controlled tasks were carried out to see to what extent morphosyntactic knowledge was present and to what extent this was influenced by task. Van de Craats sees her hypotheses confirmed: problem areas in L2 Dutch acquisition are mainly due to differences with the L1. Van de Craats proposes a more specifically focused remediation aimed at giving learners more insight into their mistakes, resulting in more efficient learning. This recommendation is illustrated through several training and practice approaches, including having learners discover the function of Dutch verb inflections, letting them seek out their own mistakes in similar sentences, and training them with semi-productive exercises. The design of such learning practices forces learners to focus on the structural difficulties and areas of stagnation caused by the differences between the L1 and L2.

Verhagen (Reference Verhagen2005) addresses the issue of a possible relationship between the acquisition of non-modal auxiliaries and the acquisition of syntactic phenomena in an L2. The data are taken from the European Science Foundation (ESF) corpus, which consists of utterances of second-language learners, and concern interview data and film retellings from two Turkish and three Moroccan learners who learned Dutch in a naturalistic setting. The data show that all five learners fail to produce topicalized structures and post-verbal negations as long as they do not produce the verb ‘to have’. Moreover, learners start to produce such verb second structures (the typical Dutch inversion of subject and verb after an introductory element in a main clause) shortly after they have produced their first instances of ‘to have’. These findings seem to indicate that the production of ‘to have’ is related to the acquisition of topicalization and post-verbal negation, i.e., the non-modal auxiliary seems to serve as a bootstrap into the V2 system of Dutch. Verhagen explains this observed relationship in the light of an explanation by Jordens (Reference Jordens2004), which suggests that L2 learners at the early stages of acquisition do not yet mark finiteness by morphological means, but rather use a class of illocutionary force markers such as hoefniet, or kanniet, and these illocutionary markers all bear a modal meaning and have scope properties in that they relate to the predicate in the sentence (e.g. ik hoefniet ziektewet ‘I need not sickness benefit’). Verhagen shows that there is both system and variation in the acquisition process of L2 Dutch: even though the acquisition order of certain verbal constructions is similar, the learners also show variation, which can largely be attributed to structural differences in the mother tongue.

Goossens (Reference Goossens2003) studies three factors that may influence the claimed efficiency of explicit learning: individual learner differences, the linguistic aspect to be learned, and the type of instruction. In her study, she investigated the short and long-term effects of explicit and implicit instruction in primary school on learning of lexical and syntactic aspects for pupils in grade 6 (11/12 year olds) in a communicative, meaningful context versus a setting without a communicative context. Progress was measured through both implicit and explicit knowledge tests. Explicit instruction, if embedded in a communicative context, turns out to be more effective than implicit instruction in a communicative context, even after one year. However, explicit instruction appears to be especially effective for introducing (new) lexical language items; for syntactic language items, the results are much less clear. The effect of explicit instruction only (without communicative context) is significantly reduced in the long term.

Bienfait (Reference Bienfait2003) reports on the results of her Ph.D. research (Bienfait Reference Bienfait2002) into the effect of explicit focus on form-focused instruction (FFI) of grammar on the development of grammatical proficiency in the Dutch of ethnic minority teenagers who are enrolled in a remedial program. Possible effects were investigated in two different functional language tasks (a formal and an informal task), and for two types of grammar structures (meaningful and less meaningful structures). Bienfait finds that the grammatical skills of pupils who received FFI improved significantly on both language tasks when they were ready for the particular grammatical structure, but pupils who were not ready did not improve, not even in the long term. However, pupils who were in a sensitive developmental stage and who received teaching focused on meaning also significantly improved their grammatical skills on both language tasks in much the same way as the students receiving FFI. An analysis of the development of individual students is not necessary, because they will learn particular structures through less structured yet meaningful input when they are ready for it. Bienfait, however, does state that explicit FFI may support the implicit acquisition of relatively complex structures, although it is doubtful whether the effect would last.

Andringa (Reference Andringa2005) also reports on the limited value of explicit instruction. According to Andringa, FFI research has tended to use explicit measures of progress, and studies that did measure implicit progress often did not compare explicit and implicit types of instruction. In a carefully controlled study, L2 students are offered implicit or explicit materials in three different grammar areas. The implicit teaching condition aims at understanding meaning. The explicit condition uses exactly the same input, but addresses the syntactic structures explicitly. The results are in line with the predominant findings in FFI research that explicit instruction is more effective in promoting L2 knowledge as measured by explicit knowledge tests, but the results provide no evidence of an interface. However, explicit instruction did promote implicit knowledge only of the simple morphological target structure if the L1 and L2 had a similar structure. For the complex syntactic structure (subordination), implicit instruction ultimately led to equal amounts of explicit knowledge. Andringa concludes that explicit instruction is more effective in promoting explicit knowledge of simple structures, and that implicit instruction promotes explicit knowledge only when the structure is complex. He also found that the individual factor developmental readiness (DR) is related to both explicit and implicit progress in the degrees of comparison. Yet, the precise involvement of DR in language learning is difficult to determine.

8. Pronunciation and gesture

Little applied research was published on the study of pronunciation over the past years. The benefits of pronunciation training are discussed by Lowie (Reference Lowie2004), who argues that high quality pronunciation training in line with the learner's goal is very useful. Reviewing previous research on pronunciation, he singles out a number of features crucial for effective training. Lowie also sees an important role for computers in pronunciation training, but most current pronunciation training programs are written from a computer science perspective rather than from a modern acquisitional or didactic viewpoint. He discusses the newly developed computer program ‘Ellips’, the result of a joint project by four universities and a software company, which works with ‘traditional’ listening and speak-and-record exercises saved in a portfolio for assessment by the teacher. The novel feature of the program is that it is based on recent insights from pronunciation education. The three steps of perception, articulation and pronunciation guide students through the practice material. For more detail on the development of ‘Ellips’, see Corda & Jager (Reference Corda and Jager2004).

Yoshioka (Reference Yoshioka2005) reminds us that gesture can be used as a useful tool to examine the internal processes of second-language use, examining how speech and gesture contribute to the creation of meaning in discourse in both L1 and L2 settings. Narratives related by Dutch learners of Japanese in a story-retelling task were video-taped and compared to the native baseline. Dutch and Japanese were chosen as they are typologically different. In their gestures, Japanese focus more on scene settings of the story, and Dutch speakers pay more attention to movement and directions. Thus, depending on the language spoken, speakers may focus differently even when providing retellings of the same story. The difference in the frequency of gesture in L1 Dutch and L1 Japanese seems to be related to the availability of linguistic means to mark the introduction and tracking of referents. The results seem to suggest that at a certain point in the development of proficiency in an L2, two seemingly different speech-associated gestures accompany reference introduction and tracking, although both are non-target-like. One type appears to be L2-specific and the other type seems to be similar to the norm of the learners' L1. Yoshioka proposes that the two seemingly different phenomena of speech-accompanying gesture may reflect interactions between the learners' L1 and L2 at two different levels: one at the level of encoding, and the other at the level of creating a semantic structure to be filtered into the target language.

9. Aptitude

Can musical talent facilitate L2 learning? Irizarri van Suchtelen (Reference van Suchtelen2005) investigated the parallels between linguistic and musical processing. An extensive exploration of linguistic and musical theory, aptitude studies, and the literature on the processing of language and music, led the author to believe that a relationship between linguistic and musical forms of aptitude could be found in aspects of working memory. Four tests were designed to measure an individual's working memory spans for linguistic and musical processing: two complex tests in which melodies and L2 sentences had to be remembered, and two non-complex tests in which tones and L1 pseudo-words had to be remembered. The results of 70 participants from a secondary school on these four tests and on a French listening examination were compared. Significant correlations between the scores on these five measures were found. Furthermore, the complexity of the information to be processed, the musical experience of the participants and other factors were found to have an effect on correlations. Especially at higher levels of processing, musical and linguistic processing capacities seem to be associated. No definitive conclusions could be drawn, among other things because of the low reliability of two of the tests, but the results do encourage further research into this relatively new area in L2 acquisition, although large-scale practical implications are less evident.

A key component of language aptitude is suggested to be working memory (e.g. Robinson Reference Robinson and Robinson2002). Havik (Reference Havik2005) measured the on-line processing of syntactic complex sentences with a self-paced reading task. Working memory (WM) capacity was measured with highly controlled reading span measures in both the L1 and the L2. The scores of 24 German L2 learners of Dutch are compared to those of a control group of 24 mother tongue speakers of Dutch. For the German learners, the scores of the State Exam of Dutch as an L2 were available. The results show that WM capacity is indeed related to both the scores on the State exam and the off-line comprehension of syntactic complex sentences in the L2. Although WM capacities in the L1 and the L2 are strongly correlated, only WM capacity in L2 is correlated to L2 reading comprehension. However, the relationship between working memory capacity and on-line sentence processing is less evident. Havik argues that this may be related to different problem solving strategies.

10. Computer Assisted Language Learning

Westhoff (Reference Westhoff, Ahlers, Kreeft, Diephuis, Leenheer and Vos2002) discusses the added value of ICT. Using ICT, language teachers can ensure input, focus on meaning, focus on form, and the training of receptive and productive strategies. He discusses the value of eight possible computer applications for foreign-language education: the computer as provider of input, means of communication, moderator of input, support in writing processes, source of information, correction instrument, practice instrument, and test instrument. After reviewing several applications such as language quests and digital classrooms, Westhoff concludes that the value of ICT comes mainly from the use of ordinary applications that have not been especially designed for education, such as the internet, text processors, checkers, etc. On the basis of his evaluation, Westhoff points out that more prototypes and examples of educational tasks and test forms need to be developed, as well as a manageable system with assessment criteria, to increase the potential value of most functions.

Reviewing the possibilities of chatting in an educational setting, Poulisse (Reference Poulisse2002) discusses several chat projects at Dutch secondary schools and presents recommendations for large-scale use of the communication medium. She points out that chatting in Dutch foreign-language education is not yet an established teaching method, but that teachers should be aware of the great potential that chatting has for language learning because of the motivating effect of real-life and functional input. Language users may be encouraged by the fact that pronunciation plays no role in chatting and that there is more time for reflection than in speaking. From the projects she analyses, Poulisse distills recommendations for effective chatting. It is important that chat groups are not too large; one-to-one sessions are more suitable for chatters who know each other. Chatting with similar proficiency levels results in more depth of conversation, and preparation is important for less proficient language users. Giving students a specific assignment such as getting information or deciding on a piece of advice makes chatting more efficient. For chats aimed at language learning, feedback sessions in which the printed chat protocol is discussed may be beneficial. However, if chat sessions are used to motivate students to use the target language or to get into contact with foreigners, feedback sessions are best left out. Also, it is important to decide whether you want students to chat with native speakers or with other FL learners. A final practical matter is where and when to chat. As both partners need to be on-line, it may be best to let the chatters themselves schedule a time to chat, or to arrange for partners or groups to chat in the school computer room at fixed hours.

The benefits of chatting for L2 Dutch learners were demonstrated by Toorenaar (Reference Toorenaar2002) who qualitatively analysed chat logs from a group of L2 Dutch learners who participated in a chat project. Like Poulisse, Toorenaar stresses the importance of monitoring and self reflection for language learning. After analysing chat logs (248 minutes) of six immigrant chat pairs using a conversational analysis paradigm, Toorenaar concludes that chatting stimulates language acquisition through the registration of the conversation on screen, the role played by time, and the lack of a shared visual context. Chatting allows L2 learners effectively to help each other in their acquisition process. However, not all chat assignments are equally effective. A good assignment ought to be realistic, have some depth and must allow for a conversation to take place. Toorenaar also stresses the importance of a feedback lesson.