Fig. 1.

Schematic of 3 possible similarity structures between a subset of the trained words. Each lozenge contains the orthographic, phonological, and semantic form of an item, with items from one orthography in blue lozenges, and those from the other in red lozenges. A thicker line between pairs indicates greater similarity. (A) Orthographic similarity reflects the number of symbols (out of 4) shared in the same position, although analyses also examined symbols shared across positions. (B) Phonological similarity reflects the number of phonemes (out of 3) shared in the same position. (C) Semantic similarity reflects shared semantic category. Note that phonological and semantic similarity analyses excluded within-orthography pairs, and so were not confounded by orthographic similarity.