Abstract The objective of this research was to further elucidate the split fovea theory. A lexical decision task was presented to central vision, while accurate fixations from participants were ensured. A monocular design was employed to try to limit factors requiring addition interpretation. The hypotheses assumed that the target word would be split by the vertical midline and initial and end letters would be projected contralaterally. This contralateral projection would be demonstrated via the disparate granularity of processing preferences displayed by the hemispheres, leaving a signature relating to the location of early visual processing. The target words were divided into four experiment conditions, varying on the level of neighbours for the initial and end letters (trigrams). A low end neighbour value was assumed to facilitate response, as the neighbourhood effect is well established, and a fine coding left hemisphere displays utility at rapidly identifying the most salient interpretation. Equally a high initial neighbour value was assumed to reduce latency as right hemisphere coarse coding will display a preference for a broad range of activations. Fixation locations were controlled for, and analysis conducted relating to the location of fixation. Fixation position generally displayed no major influence on the discovered interactions. Sex based difference were also monitored for, with some minor affects observed. The main result was unpredicted. Conditions whereby response for one hemisphere was facilitated produced notably shorter latencies. The condition where both hemisphere were facilitated showed no advantage for response latency, and latencies were comparable with the condition in which neither hemisphere was facilitated. These unpredicted results favour a new model of hemispheric control; words which allow a clean predictable seizure of hemispheric control demonstrate the largest facilitation. This proposed Hemispheric Control Model (HCM) also accounts for slight sex-based differences observed. A HCM would require further replication, and if the model proves robust, application to binocular and continuous reading paradigms.