Differences in Children’s Spelling Performance for Typed and Handwritten Dictations

Literacy acquisition is essential to be able to participate in our knowledge society. Reading and writing are learned skills that children acquire after intensive practice. So far, learning to write has been mainly based on practice that relies on handwriting. However, the increasing digitalization and the resulting availability of devices that require us to type instead of writing by hand, have raised questions as to whether typing should take a more prominent role in spelling and writing education. In addition, discussions about optimal practices for testing spelling and writing skills have also pointed to the potential role and possible advantages of digitization at various levels, from employing text-to-speech technology (TTS) for replacing the reading of the dictation by teachers to providing automatic feedback on spelling errors.

So far, little research has addressed the impact of digitized dictation on spelling and writing skills, mainly due to the limited availability and implementation of digital dictation software in mainstream writing instruction, especially for languages other than English. On the other hand, recent attempts at digitizing at least some aspects of this process have produced valuable data sets that can be employed to start this line of research.

One such example is the Dutch BasiScript corpus, which contains written texts by Dutch primary school children (grades 2-6) collected through a traditional form of dictation in which the teacher read the word in context of a sentence and the children wrote it down by hand. All written texts were manually digitized and made available as is, including metadata about the children and the schools. Another relevant corpus was collected by Zwijsen Publishers through digital testing software. The dictation sentence was read by a pre-recorded voice and primary school children (grades 2-6) had to type one specific word from the sentence. In these two corpora, different materials were employed for dictation, but an overlap of twenty words asked in the same grade was observed. In the current research we employed this dataset for a first, exploratory investigation to the effects of typing and handwriting on Dutch spelling performance by addressing the following research question: To what extent do typing and handwriting lead to differences in Dutch spelling performance?

To answer this research question, we analysed the overlapping words in the BasiScript and Zwijsen corpora in terms of unique spelling variants (variants of the word the children were asked to spell) and relative error frequency (percentage of misspellings). We found significantly more unique spelling variants in the Zwijsen dataset compared to BasiScript dataset, while the relative error frequency was found to be significantly lower in Zwijsen than in BasiScript. So, we can conclude that typing and handwriting do indeed lead to differences in Dutch spelling performance.

As the corpora were collected in different ways, many variables were uncontrolled for, and since the dataset was limited, these preliminary results should be treated with caution. Additional research is necessary to gain a better understanding of how different writing modalities affect spelling performance.