Literacy education in secondary schools is experienced as dull and ineffective. We investigate how reading and writing education in lower grades of vmbo-tl and havo can be made more challenging and meaningful through a cross-curricular genre-pedagogical approach. Students learn to deliberately apply genre knowledge in Dutch and history lessons. With teachers we design lesson units for deliberate genre literacy. Effectiveness is tested in a large-scale intervention study. A cross-curricular genre pedagogy is expected to increase students’ knowledge, skills and motivation. We also investigate what subject-specific didactic expertise is needed for teachers, and what it yields for integration between curriculum subjects.

• dr. Erwin Mantingh
• dr. Jacqueline Evers-Vermeul
• dr. Renske Bouwer

• dr. Ninke Stukker (Rijksuniversiteit Groningen)
• dr. Jannet van Drie (Universiteit van Amsterdam)
• prof. dr. Carla van Boxtel (Universiteit van Amsterdam)
• dr. Jeroen Steenbakkers (Hogeschool Windesheim)
• Stefan Glasbergen (Hogeschool van Amsterdam)
• prof. dr. em. Kees de Glopper (Rijksuniversiteit Groningen)
• prof. dr. Wilbert Spooren (Radboud Universiteit Nijmegen)

EU-COST program for academic exchange

Effective discourse in any language is characterized by clear relations between sentences and coherent structure. But languages vary in how relations and structure are signalled. While monolingual dictionaries and grammars can characterise the words and sentences of a language and bilingual dictionaries can do the same between languages, there is nothing similar for discourse. For discourse, however, discourse-annotated corpora are becoming available in individual languages. The project will facilitate European multilingualism by (1) identifying and creating a portal into such resources within Europe - including annotation tools, search tools, and discourse-annotated corpora; (2) delineating the dimensions and properties of discourse annotation across corpora; (3) organising these properties into a sharable taxonomy; (4) encouraging the use of this taxonomy in subsequent discourse annotation and in cross-lingual search and studies of devices that relate and structure discourse; and (5) promoting use of the portal, its resources and sharable taxonomy. With partners from across Europe, TextLink will unify numerous but scattered linguistic resources on discourse structure. With its resources searchable by form and/or meaning and a source of valuable correspondences, TextLink will enhance the experience and performance of human translators, lexicographers, language technology and language learners alike.

• dr. Jacqueline Evers-Vermeul
• Jet Hoek MA

State-of-the-art machine translation (MT) systems, especially statistical but also rule-based
ones, operate in a sentence-by-sentence mode, and do not propagate information through the series of sentences that constitute texts. Such a propagation is however helpful, and sometimes even indispensable, to make correct translation choices for words and phrases that depend on previous ones. The goal of MODERN is to model and automatically detect such dependencies, and to study their integration within MT, with the aim of demonstrating improvement in translation quality.
The focus of MODERN is on the interplay between referring expressions such as noun
phrases and pronouns, which must be coherently translated throughout a text, and discourse relations between sentences, which are often conveyed by explicit connectives that are notoriously difficult to translate. MODERN will study joint computational models of discourse entities and discourse relations in texts, based on linguistic theories and experimental grounding, and the inclusion in such models of automatically generated domain-knowledge related to the discourse entities. MODERN will design and implement these probabilistic models, and integrate them with operational MT systems, both rule-based (Apertium) and statistical (Moses).
Particular attention will be paid to the evaluation of MT improvement, studying the eect on
human readers of various translation options for discourse entities and connectives, and aiming to optimize MT output in this respect. The MODERN project will focus on four languages {English, French, German and Dutch} for which the partners have considerable expertise. Two domains will be used as case studies: Alpine texts from a multilingual corpus of Alpine Club yearbooks (Text+Berg) and texts on environmental legislation and debates extracted from the JRC-Acquis, DGT-Acquis, and Europarl parallel corpora.

• Jet Hoek MA
• dr. Jacqueline Evers-Vermeul

While there is an urgent need for robust readability assessment tools, applicable to a range of communicative domains, none of the existing tools for Dutch offers a valid empirical basis and sufficient functionality. Recent developments in computational linguistics and discourse processing create possibilities to change this situation. We propose to develop a validated reading level tool, aiming at secondary school readers, adult readers of public information, and by extension, the Dutch-speaking population at large. First, a text-analytic tool is developed using the latest results of computational linguistics research. Second, cloze comprehension data are collected among secondary school readers, in a design assessing both differences between texts and between versions of the same text. Third, a subsample of texts is re-used to investigate on-line processing in eye-tracking studies. The combination of comprehension data and on-line processing measures provides insight in the way textual features affect the construction of cognitive representations. Fourth, additional cloze data are collected for adults reading government-citizen information texts in The Netherlands and Flanders. Fifth, the relation between cloze data and reading times on the one hand, and text features on the other hand, is analyzed in a multilevel regression analysis and in a machine learning study. These analyses are used to develop a reading level prediction tool: LIN (LeesbaarheidsIndex voor het Nederlands). This validated reading tool is relevant to various domains in society: education, publishing, government-citizen communication. It will provide the foundation for developing domain-specific readability and writing tools that has been missing so far.

• dr. Suzanne Kleijn
• dr. Henk Pander Maat

AIM OF THE PROJECT

Bilingual children grow up with two languages. Therefore, they cannot hear and speak each of their languages as often as monolingual children do. As a result, bilinguals tend to develop their languages slower and are often referred to speech language therapists for screening and intervention. Due to the lack of reliable assessment tools for bilinguals, these children are often misdiagnosed for language impairment, which leads to unnecessary treatments and emotional problems. Alternatively, bilingual children can be under-diagnosed, because speech language therapists may think that their language delay is due to bilingualism, rather than language impairment. In this case, children do not receive speech therapy that they really need.

Hence, insights and tools for differentiating between typically developing bilinguals and children with language impairment are urgently needed. This is exactly what our project set out to achieve. In this study, we compared language production and comprehension by bilinguals with typical language development and monolinguals with and without language impairment. Our main goal was to find a way (e.g. tool, method, analysis) to distinguish between typically developing bilingual children (who may just be a bit slower due to dual language exposure) from bilingual children suffering from a language disorder.

MAIN FINDINGS

Our main and most important finding is that there is a way to distinguish between typically developing bilinguals and children with language impairment. The two groups look very similar in language production, but have clearly different profiles in language processing.

When we compare the narratives produced by bilingual children (in their weaker language) to narratives produced by monolinguals with language impairment, we see a lot of similarities. For example, both groups often use the wrong conjunctions to connect clauses (e.g. ‘and’ instead of ‘but’). These errors are numerous until age six, but are also present in the speech of children as old as eight years of age. Sometimes bilingual children make even more errors than children with language impairment; this is particularly evident in the domain of pronominal gender. Hence, based on errors in speech production it is difficult (of not impossible) to say whether a bilingual child suffers from a language disorder or not. This finding is consonant with the observation that bilingual children are often misdiagnosed or under-diagnosed for language impairment.

However, two distinct profiles emerged when we looked at online comprehension measures (by means of eye-tracking). Our results reveal that children with language impairment have difficulty understanding discourse connectives and are not able to predict discourse continuation based on connective semantics. As against this, bilingual children perform as well as typically-developing monolinguals: their gaze patterns clearly show that they know subtle semantic differences between complex discourse connectives and use this information to predict how discourse will unfold. The same pattern emerged in the domain of pronominal gender. Bilingual children under age 6 make a lot of errors in pronoun use, for example saying ‘he’ instead of ‘she’. However, bilinguals of this age do show sensitivity to gender cues in receptive tasks and look at the picture congruent with the gender of the pronoun, just like typically developing monolingual children of their age.

Evidence from our eye-tracking studies suggests that the seemingly similar profiles of bilinguals and children with language impairment (in speech production) probably have different underlying causes. Children with language impairment make errors in their speech because they have difficulty learning the semantics and the grammar of discourse connectors. In contrast, typically developing bilinguals make errors under the influence of their dominant language, but they do know the correct word meanings and grammar rules. In other words, bilingual children sometimes have difficulty inhibiting their dominant language when they are speaking. The capacity to inhibit the dominant response improves with age, as a function of cognitive maturation.

Another crucial finding from this project is that bilingualism does not necessarily lead to problems and language delays. When we look at the dominant language of our bilingual participants – German in Germany and Dutch in the Netherlands – we do not see many differences from monolingual German- and Dutch-speaking children (even though there is some evidence of crosslinguistic influence in the bilingual mind). However, this only holds for children growing up with two languages from birth; children acquiring German or Dutch as their second language do feature some delay and/or deviations in their linguistic development. Furthermore, bilingual children have difficulty with their non-dominant language, that is the language that is not spoken or maintained in the country of residence. Since harmonious bilingual development is essential for a child’s well-being, it is very important to support the acquisition of minority languages, for example, by creating more bilingual schools and kindergartens.

An important implication of this research is that speech language therapists should be aware of the fact that language production of typically developing bilinguals often looks very similar to that of individuals with language impairment and that (online) comprehension measures provide a much better tool for differentiating between the two groups than language production measures (such as narratives).

This project was supported by the EU grant (Marie Curie IRSES) and was carried out in cooperation with Zentrum für Allgemeine Sprachwissenschaft (Berlin) and Herzen State Pedagogical University of Russia (Saint-Petersburg).

• prof. dr. Sergey Avrutin
• dr. Pim Mak
• dr. Elena Tribushinina

Although discourse is a crucial level in language and communication, existing corpora of Dutch langauge lack annotation at this level. The DiscAn project sets the first step to change this situation for Dutch, in line with international tendencies. It has five main goals: 1) to standardize and open up an existing set of Dutch corpus analyses of coherence relations and discourse connectives; 2) to develop the foundations for a discourse annotation system that can be used in Dutch natural language corpora; 3) to improve the metadata within CLARIN by investigating existing CMDI profiles or adding a new CMDI profile specially suited for this type of analysis; 4) to inventorize the required discourse categories and investigate to what extent these could be included in ISOcat categories for discourse that are currently being developed; 5) to further develop an interdisciplinary discourse community of linguists, corpus and computational linguists in The Netherlands and Belgium, in order to initiate further research on cross-linguistic comparison in a European context.

De hoofdvragen van dit onderzoek zijn: Hoe verloopt het leesproces van vmbo’ers bij het lezen van verhalende teksten en studieboekteksen en hoe kunnen we dat leesproces vergemakkelijken? Het vmbo is de grootste onderwijssector binnen het voortgezet onderwijs. Hacquebord (2007) laat zien dat een kwart van de vmbo’ers niet in staat is de studieboekteksten te begrijpen.

Land (2009) toont aan dat vmbo’ers lezen ook niet leuk vinden en de studieboeken saai vinden. Land laat daarnaast zien dat bepaalde tekstkenmerken van invloed zijn op de begrijpelijkheid en de waardering van studieboekteksten. De vmbo’ers blijken ten eerste profijt te hebben van coherentie. Leerlingen begrijpen een tekst beter wanneer de structuur van de tekst expliciet is gemarkeerd dan wanneer deze impliciet is gelaten. Ten tweede blijken vmbo’ers identificerende studieteksten meer te waarderen, maar die teksten leveren lagere scores op tekstbegriptoetsen op dan distantiërende teksten.

Het offline-onderzoek van Land roept allerlei vragen op. Wordt de identificerende tekst anders verwerkt dan de distantiërende tekst? Leidt een identificerende tekst de aandacht af van de belangrijke informatie? Gebruiken betere lezers andere leesstrategieën dan zwakke lezers? Allemaal vragen die nog niet zijn beantwoord.

Het huidige project probeert deze leemte te vullen. Op grond van de resultaten van procesonderzoek kan worden nagegaan wat de theoretische verklaring is voor de gevonden effecten onder vmbo’ers. De praktische toepassing is dat we adviezen kunnen opstellen hoe onderwijsteksten te formuleren en kunnen zwakke lezers geïnstrueerd worden hoe ze een tekst moeten lezen. Coherentiemarkering (expliciet versus impliciet) en identificatie (identificerend versus distantiërend) zullen in dit onderzoek de twee tekstvariabelen zijn waarvan we het effect op het leesproces van vmbo’ers nagaan. Ook het type lezer (zwakke versus sterke lezers) en het genre (zakelijke versus narratieve teksten) zullen in ons onderzoeksdesign worden opgenomen.

• dr. Jacqueline Evers-Vermeul

Leesvaardigheid speelt een cruciale rol in de schoolcarrière van middelbare scholieren. Om te kunnen slagen moeten leerlingen immers studieboekteksten lezen, begrijpen, leren en de leerstof reproduceren tijdens examens of toetsen. Er zijn echter veel leerlingen voor wie het goed kunnen lezen niet vanzelfsprekend is. Vooral in het voorbereidend middelbaar beroepsonderwijs vormt de leesvaardigheid een groot probleem. In dit project gaan we de effecten van zowel bepaalde lezerskenmerken als bepaalde tekstkenmerken op de leesprestatie na. Hierbij wordt gericht op leerlingen van het vmbo die vaak getypeerd worden als zwakke lezers, en de structuur- en stijlkenmerken die voorkomen in de studieboekteksten van deze leerlingen. Hierbij staan de volgende vragen centraal:

- In hoeverre hebben structuur- en stijlkenmerken effect op de mate waarin vmbo-leerlingen leerteksten begrijpen en waarderen?
- In hoeverre hebben lezerskenmerken effect op de mate waarin vmbo-leerlingen leerteksten begrijpen en waarderen?

Dit project wordt gecontinueerd in het project "De verwerking van verhalende en studieboekteksten door vmbo’ers".

Causal relations between discourse segments can be expressed by connectives like because and so. What does it mean when speakers use one of two similar connectives rather than the other? How do language users process causal relations? And how do children acquire these textual building stones?

These are important, essentially unanswered questions about discourse, a key component of human communication. This program studies human cognition by investigating the mechanisms underlying discourse coherence. Starting from the challenging idea of a direct link between linguistic categorization and cognition, causal connectives are investigated. Recently established (text-) linguistic insights suggest Causality and Subjectivity are salient categorizing principles.

The central hypothesis is that, together, these principles account for causal coherence and connective use, and play a pivotal role in explaining cognitive complexity in discourse. This hypothesis is tested in three subprojects, investigating:
1. connectives in spoken and written discourse in Dutch, German and English
2. the acquisition of connectives and
3. on-line discourse processing.

Causality is considered a universal principle, which is systematically encoded in language. The way Subjectivity - distinguishing between causality residing in the world or in the speakers mind - cuts up the causal lexicon, is expected to vary across languages and (spoken versus written) modalities. Causality and Subjectivity are hypothesized to determine the complexity of causal relations are considered more complex than additice relations. This inherent complexity should account for connective acquisition - rather than parental input does - and discourse processing - rather than schematic expectations do. Using the innovative, multi-disciplinary methodology of concerging evidence, the program integrates (sub) disciplines that have hardly been related: (text-)linguistics, analysis of spoken and written corpora, language acquisistion and discourse processing. Results are likely to clarify previously unsolved issues in language use, which implies a significant contribution to a Cognitive Theory of Discourse Representation.

This project is currently being continued in other projects.

Keywords: Causality, Subjectivity, Connectives, Discourse processing, Language Acquisition

• dr. Jacqueline Evers-Vermeul

• dr. Henk Pander Maat

Everyday, digital media play a more important role in our society and in children’s lives. In the Netherlands, practically all children are online nowadays. Besides playing games, children use digital media for social gatherings, to create their own digital space and to find information as a support for learning.
Researchers report all kind of problems children encounter during information-seeking, because they are confronted with information systems that are designed by and for adults. Therefore, the last decade, many digital environments have been developed with a child-friendly interface, especially for children. Several questions arise from these developments, such as: do these search environments really support children in effective information-seeking? And are these systems consistent with children’s cognitive needs and skills? What are the principles underlying children's search behaviour and how can we design search interfaces that are consistent with children's needs and skills and support them in effective information-seeking.

• prof. dr. Leo Lentz