Computational analysis compares how three phonics programs support word decoding

Phonics instruction, which teaches children the correspondences between written letters and spoken sounds, is widely recognized as an effective method of reading instruction. Yet the programs that deliver this instruction vary considerably in how many and which spelling-sound rules they teach, and the consequences of those differences for children’s early reading have been poorly understood. A new computational study set out to answer a deceptively simple question: what exactly do different phonics programs teach, and how well does that content allow a child to pronounce written words?

The study, published in Cognitive Research: Principles and Implications, was conducted by Hannah S. Rognan, Tobias Ungerer, Irys-Amélie Champagne, Yiannie Lin, Kathleen Rastle and Blair C. Armstrong. The team compared three widely used English-language phonics programs: Jolly Phonics, Letters and Sounds, and Read Write Inc. The benchmark was a naturalistic child vocabulary, drawn from frequent words in child-directed texts, and the measure was decodability, meaning how many of those words a learner could pronounce using only the grapheme-phoneme correspondences, or GPCs, taught by each program.

Decodability was measured through an algorithm that simulates how many words can be decoded with each GPC taught. Rather than relying on a single metric, the researchers quantified decodability in different ways to probe the potential impact of teaching GPCs that create ambiguity. Ambiguity arises in two ways: when the same letter or letters map to different phonemes, and when a more complex grapheme involving several letters subsumes letters that would otherwise be learned through simpler GPCs.

The central finding is that teaching more GPCs increases the number of words that can be decoded, but there are diminishing returns to teaching more rules once the cost in instructional time is taken into account. A program that adds rule after rule does not keep unlocking new vocabulary at the same rate; each additional correspondence helps fewer and fewer new words become pronounceable.

The analysis also uncovered a subtler trade-off. Teaching larger sets of GPCs can introduce correspondences that create ambiguity, and empirical studies show that such ambiguity can negatively impact learning. In other words, a program that looks more comprehensive on paper may actually complicate the learning process, because children must navigate cases where identical spellings correspond to different sounds or where multi-letter graphemes overlap with simpler letter-sound rules they have already been taught.

Applying these measures, the researchers found that the three programs differ in how many words they allow learners to pronounce, and in whether those words can be pronounced unambiguously. The differences among Jolly Phonics, Letters and Sounds and Read Write Inc. therefore reflect genuine trade-offs in program design rather than trivial variations in presentation, with practical constraints such as classroom time shaping what any program can realistically deliver.

Beyond comparing existing programs, the team used their algorithm prospectively, to identify additional GPCs that could improve the potential effectiveness of the programs they analyzed. This turns the tool from a descriptive audit into a design instrument: instead of asking only how well a given curriculum performs, researchers and curriculum developers can ask which specific correspondences, if added, would most increase the number of words children can decode.

The authors frame the work as providing targeted guidance on the factors that need to be considered in designing an optimized phonics program. They also position the algorithm itself as a practical contribution, one that can facilitate phonics program design and pinpoint the empirical research still needed to advance toward an optimized set of GPCs. Because the study is a computational simulation of decodability rather than a classroom trial, its conclusions concern the potential of program content to support pronunciation, not directly measured effects on children’s reading outcomes.

The significance statement accompanying the paper notes that children’s knowledge of phonics, the systematic letter-sound correspondences taught during the initial years of schooling, is one of the strongest predictors of their later reading ability. The authors suggest that the features of different programs trade off in relation to practical constraints in the classroom, and that their approach can inform the design and assessment of phonics programs and other educational materials, as well as the selection of effective teaching resources by teachers and school boards.

For educators, the practical message is that the number of rules taught is not the only variable that matters. The specific correspondences chosen, the ambiguities they introduce, and the instructional time they require all interact to determine how much of a child’s vocabulary a program makes pronounceable. The study’s open-access publication means that the analysis and its guidance are available to the teachers, school boards and researchers who ultimately decide which letter-sound rules reach the classroom.

Subject of Research: Computational comparison of the grapheme-phoneme correspondences taught by three English phonics programs and their support for word decoding

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

Article References: Rognan, H. S., Ungerer, T., Champagne, I.-A., Lin, Y., Rastle, K., & Armstrong, B. C. (2026). How do different phonics programs support word decoding? Insights from computational analyses. Cognitive Research: Principles and Implications. https://doi.org/10.1186/s41235-026-00753-7

Image Credits: AI Generated

DOI: 10.1186/s41235-026-00753-7

Keywords: phonics, grapheme-phoneme correspondences, word decoding, reading instruction, Jolly Phonics, Letters and Sounds, Read Write Inc, decodability, computational modeling, literacy education, child vocabulary, reading development