Early in children’s schooling, they are faced with the challenge of learning to read. One of the most important early skills that children must master is “decoding”, the ability to map spelling to sound, e.g., to pronounce a string of letters. A popular model of reading acquisition, “triangle model” (Seidenberg, 2005) suggests that children begin learning to read via decoding: they link orthography to phonology (the sound pattern of a word) and can then use the intact phonology-semantics system (e.g.spoken word recognition and production) before reaching the adult-like ability of going directly from orthography to meaning (e.g. “sight word reading”).

An illustration of the "triangle" model in which three cognitive facilities interact to enable various language skills.

Thus, decoding sets the early stages for much more fluent reading later, and a wealth of evidence attests to its importance (Erhi, Nunes, Stahl & Willows, 2001). However, these mappings between sound and spelling are complex. We like to think of them in terms of “rules” (e.g., use the long “I” when you see EA, as in BEAD and NEAT), but in actuality, there are lots of exceptions to the rules (e.g., HEAD, LEAD, BEAR), and it might be more useful to think of them in terms of more graded and probabilistic relationships, quasi-regularities.

In collaboration with a local company that develops reading intervention software (Foundations in Learning), and with the West Des Moines Community schools, we are investigating how children learn decoding skills ask several major questions about this developmental process:

  • How do children harness the statistics of the orthographic system to learn new mappings?
  • What role does variability in word forms play in guiding children’s learning?
  • Can we independently children’s skills along the separate pathways between orthographic, phonological and semantic information?
  • What is the best way to structure a program aimed at helping children learn to read?

We implement these studies directly in the student’s classrooms using computer-based educational software. This allows us very precise control over the stimuli and the experience of the children, that we need for isolating learning mechanisms. However, by using an existing reading intervention, the principles that we uncover can be translated directly to interventions and curricula.


Seidenberg, M. S. (2005). Connectionist models of word reading. Current Directions in Psychological Science, 14, 238-242.

Ehri, L. C., Nunes, S. R., Stahl, S., & Willows, D. (2001). Systematic Phonics Instruction Helps Students Learn to Read: Evidence from the National Reading Panel's Meta-Analysis. . Review of Educational Research, 71(3), 393-447.