Date of Degree


Document Type


Degree Name



Educational Psychology


Linnea Ehri

Committee Members

Alpana Bhattacharya

Katharine Pace Miles

Subject Categories

Curriculum and Instruction | Early Childhood Education | Education | Elementary Education | Language and Literacy Education


Phonics, Reading, Synthetic Phonics, Decode, Blending, Segmentation


Two methods of decoding instruction were compared. Kindergartners who could not decode nonwords participated in the study, N = 38, M = 5.6 years. Segmented phonation, frequently used in synthetic phonics programs, taught students to convert graphemes to phonemes by breaking the speech stream (“sss – aaa – nnn”) before blending. Connected phonation taught students to pronounce phonemes without breaking the speech stream (“sssaaannn”) before blending. Kindergartners were matched and randomly assigned to the two conditions. Both groups were taught to decode the same set of CVC nonwords consisting of continuant consonants and vowels that could be stretched and connected without altering their phonemic identities and without breaking the speech stream. Following learning, students completed a transfer task to decode CVC nonwords with stop consonants that are harder to blend because a schwa vowel is added when stops are pronounced separately (e.g., “dǝ – æ – pǝ”) and must be deleted during blending (e.g., “dap”). It was hypothesized that the connected phonation group would better understand how to blend phonemes than the segmented phonation group and hence would be able to overlook schwa vowels more easily in the transfer task. Results were supportive. Connected phonation students read the CVC nonwords with stops more accurately on the transfer task as well as a delayed posttest. Errors showed that segmented phonation students more often forgot the phonemes they had pronounced, especially initial phonemes, when they tried to blend them. In addition to treatment effects, children’s sight word reading explained significant variance in nonword reading on the transfer task. Findings carry implications for how to teach decoding more effectively.