Forty kindergarten children participated in a study to determine (1) whether significant relationships existed among children's scores on various word boundary tasks, (2) whether scores on the individual tasks were significantly related to the children's ability to represent phonemic segments in their spelling, and (3) whether different instructional approaches made a difference in children's recognition of word boundaries and phoneme segmentation ability. The children came from classes taught by two different teachers--one who used no specific reading instruction approach, and one who used a structured, synthetic phonics approach. Each child was given four word boundary tasks and a spelling test. The children's performance on the four tasks differed significantly across teachers. For students in the "no phonics" classes there were significant correlations among all four tasks; in addition, performance on each task was significantly related to the children's ability to represent phonemic segments in their spellings. (FL)