The present study investigates differences in the word-reading process between individuals reading in a deep (unpointed Hebrew) and a shallow orthography (Turkish). The participants were 120 students evenly and randomly recruited from three levels of education (primary = 3rd-4th graders; middle = 6th-7th graders; high = 9th-10th graders). The students were tested with a computerised paradigm that assessed their efficiency in determining the identicalness of real word (RW) pairs and nonsense word (NW) pairs under perceptual and conceptual conditions. Based on a strong orthographic depth hypothesis, Turkish readers were hypothesised to manifest superior word-processing skills in comparison to Hebrew readers, both for RWs and NWs. Evidence obtained from the analysis of the quantitative and qualitative performance of the participants failed to support this prediction. Findings are discussed with reference to a single-route grain-size-based word-reading model and a modified dual-route word-reading model.