As a cornerstone of education and social participation, reading develops most successfully through early instruction in classrooms that provide meaningful exposure, responsive feedback, and equitable learning opportunities.
A growing body of work reports robust associations between rhythmic skills and reading, particularly in young children and in dyslexia.
However, both rhythm and reading are multidimensional, and it remains unclear which rhythmic components align with specific reading outcomes and whether the relation is partly causal.
Motivated by these gaps, this thesis aimed to (i) develop a classroom-deployable platform that supports scalable, high-quality research in schools and potential integration, (ii) clarify component-level relations between rhythm and reading subskills, and (iii) test whether brief, classroom-deliverable training can produce measurable gains, following a Design-Based Research approach.
We designed and developed Rhythmico, a classroom-ready platform integrating digital assessment, rhythmic training, and data capture.
Across iterative deployments in grade 1 and grade 2 classrooms (total n=49), the platform supported feasible procedures, sustained engagement, and reliable measurement, all for school use.
Using this platform, we then empirically explored the link between reading and rhythm. Specifically, we focused on two components of rhythm.
Beat-based (BB) refers to the ability to entrain and synchronize to a regular pulse.
Sequence-memory-based (SMB) refers to the ability to encode and reproduce temporally ordered patterns.
In Study 1 (n=46; grades 1--2), we mapped short-term associations between rhythmic skills and reading-related outcomes. BB skills were most consistently associated with phonological awareness and rapid automatized naming. SMB skills were most consistently associated with reading fluency, word-level reading, and phonological short-term memory.
In Study 2 (n=15, grade 1), we compared a multi-week digital rhythmic intervention with a motor-control one. Both arms showed comparable improvements in several reading-related outcomes at the tested dosage, limiting strong causal inference.
In Studies 3 and 4 (respectively n=15 and n=33, grade 1), we compared brief BB-focused and SMB-focused training blocks. Study3 used different game sets, whereas Study4 held game content constant and varied only interaction style, to better control for engagement-related differences. Across these studies, BB performance improved reliably over short windows and showed alignment with reading fluency and rapid automatized naming. In contrast, SMB performance aligned more with phonological short-term memory and word-level reading, but exhibited slower short-term change. Between-condition differences in reading outcomes were small and imprecise at the tested dosage, and average change--change relations were modest, with stronger indications among weaker readers.
Overall, the findings support a component-level acc
