Abstract In alphabetic writing systems like English or French, many words are composed of more letters than phonemes (e.g. BEACH is composed of five letters and three phonemes, i.e. /biJ/). This is due to the presence of higher order graphemes, that is, groups of letters that map into a single phoneme (e.g. EA and CH in BEACH map into the single phonemes /i/ and /J/, respectively). The present study investigated the potential role of these subsyllabic components for the visual recognition of words in a perceptual identification task. In Experiment 1, we manipulated the number of phonemes in monosyllabic, low frequency, five-letter, English words, and found that identification times were longer for words with a small number of phonemes than for words with a large number of phonemes. In Experiment 2, this `phoneme effect' was replicated in French for low frequency, but not for high frequency, monosyllabic words. These results suggest that subsyllabic components, also referred to as functional orthographic units, play a crucial role as elementary building blocks of visual word recognition.