In order to become literate, a child does not only need to master grammar and different linguistic registers, but also learn to spell orthographically correct. In this part of the project, we are concerned with the question of how orthography is acquired in German schools. We want to establish which linguistic units and statistical properties of words influence the writing process of adults and children. For instance, Kandel et al. (2011) have shown that in French, children’s writing is structured by syllables. Later, however, as children become more proficient, bigram frequency (the frequency of co-occurrence of two graphemes) plays a larger role as a planning unit. This shows that writers are sensitive to such statistical properties and that they use them for orientation in their handwriting process. We assume that written word production that is rooted in such implicit orthographic knowledge will lead to fast and correct, i.e. proficient, spelling skills.
Our first aim is to assess the hypothesis that the encoding processes seen in the spelling of fluent and correct writers correlate strongly with the statistical surface properties of written language but that this correlation is less apparent in less proficient writers. If it holds, we should see that implicit grammatic and orthographic knowledge impacts differently the spelling process of spellers of different proficiency levels, specifically on the kinds of planning units used in encoding written language. By testing children and adults who have been instructed to read and write through different didactic concepts, we want to compare implicit orthographic knowledge and proficiency between these concepts. Ideally our findings can form the basis for adapting orthography instruction in schools, such that more children are enabled to master orthography than it is the case today.
We carried out three experiments with four groups of participants each – 3rd and 4th graders, 5th and 6th graders, young adults and older adults. We reorded their movements while they wrote words on a graphic tablet in a copying task. Per participant and word, we established their writing speed per letter and the pauses elapsing between the motor movements pertaining to individual letters using Ductus (Guinet and Kandel, 2010). In order to code the data for analysis, we segmented and analysed each letter a person had written. Unlike prior studies, which had focused on certain letters of intrest, we coded all letters of all words, as we asssumed that differences in spelling patterns might affect nonadjacent letters, too. This was extremely time-consuming. The data are now fully coded and we hope to submit the publications about our results by the end of 2022.