How do different phonics programs support word decoding? Insights from computational analyses

Abstract Phonics instruction involves teaching grapheme-phoneme correspondences (GPCs) and is widely recognized as an effective method of reading instruction. We examined what GPCs phonics programs teach, how those GPCs support a child’s ability to pronounce written words to understand how to create an optimized phonics program. We compared three phonics programs, Jolly Phonics , Letters and Sounds , and Read Write Inc. , in terms of how well their GPCs enabled the decoding (pronunciation) of a naturalistic child vocabulary. Decodability was measured through an algorithm that simulates how many words can be decoded with each GPC taught. We quantified decodability in different ways to probe the potential impact of teaching GPCs that create ambiguity, either because the same letter(s) map to different phonemes, or because a more complex grapheme involving several letters subsumes letter(s) from simpler GPCs. Teaching more GPCs increases the number of words that can be decoded, although there are diminishing returns to teaching more rules given the cost in terms of instructional time. Moreover, teaching larger sets of GPCs can introduce GPCs that create ambiguity, which empirical studies show can negatively impact learning. We also use our algorithm to identify additional GPCs that could improve the potential effectiveness of existing programs. Our work provides targeted guidance pertaining to the factors that need to be considered in designing an optimized phonics program. It also offers an algorithm that can facilitate phonics program design and pinpoints empirical research needed to advance toward an optimized set of GPCs.