What Do Artificial Orthography Learning Tasks Actually Measure? Correlations Within and Across Tasks.

Xenia Schmalz,Elisabetta De Simone,Kristina Moll,Gerd Schulte-Körne

doi:10.5334/joc.144

Xenia Schmalz, Elisabetta De Simone + Show 2 more

Open Access

https://doi.org/10.5334/joc.144

Copy DOI

Abstract

Artificial Orthography Learning (AOL) may act as a possible candidate to model the learning of print-to-speech correspondences. In order to serve as an adequate task, however, we need to establish whether AOL can be reliably measured. In the current study, we report the correlations between the learning of two different artificial orthographies by the same 55 participants. We also explore the correlation between AOL skill and other participant-level variables, namely Paired Associate Learning (PAL) performance, word and nonword reading ability, and age. We find high correlations between learning of two different artificial orthographies. Correlations with reading fluency and PAL are low. These results leave questions about the link between reading acquisition and AOL. At the same time, they show that AOL ability can be reliably measured and justify its use for future studies.

Highlights

Reading research often aims to assess which cognitive processes differ across individuals varying in reading ability, or how learning to read is affected by reading instruction and various
In order to explore the correlations between the reading and learning tasks, the original plan was to correlate the number of run-throughs which participants needed to learn the materials to criterion against each other and against reading ability, as this was the dependent variable used by Taylor et al (2011)
This suggests that training performance is relatively stable across time, validating the use of this task as a reliable indicator of a participant characteristic. This suggests that, from a methodological perspective, the Artificial Orthography Learning (AOL) task is well-suited to studying individual differences and behavioural correlates of this task. This conclusion is limited to the AOL paradigm as implemented in the current study: other studies have used different designs, for example, by teaching participants individual symbolsound correspondences rather than encouraging them to extract these from the presentation of whole word forms (Aravena et al, 2013; Aravena et al, 2017; Law et al, 2018), or teaching participants to read words that exist in their native language, written in a novel script (Taylor et al, 2017; Yoncheva et al, 2010)

Summary

Introduction

Reading research often aims to assess which cognitive processes differ across individuals varying in reading ability, or how learning to read is affected by reading instruction and various. According to theories of reading acquisition, the learning of print-to-speech correspondences is a sine qua non: once children learn how to derive a phonological output from an unfamiliar written word, they can use this decoding process to read almost any word that they have not seen in its written form before (Castles, Rastle, & Nation, 2018; Share, 1995). The importance of this decoding process is further demonstrated by the finding that children with dyslexia show disproportional difficulties with pseudoword reading, which is an indicator of decoding skills (Rack, Snowling, & Olson, 1992; but see Van den Broeck & Geudens, 2012). It is informative to study effects of participant-level characteristics on reading acquisition: in a natural setting, one does not know how participant-level characteristics may have affected the prior knowledge or cognitive skills that are specific to print-to-speech correspondence learning

Objectives

Results

Conclusion