talianL1 speakers of EnglishL2 produce the same English sound as longer if spelled with two than with one letter, following Italian grapheme-phoneme conversion rules. Do Italian listeners perceive short and long sounds in English homophonic word pairs that are spelled with a single letter or a digraph (finish-Finnish; morning-mourning)? In Experiment 1, 50 ItalianL1-EnglishL2 bilinguals and 50 English controls performed a Consonant Perception Task and a Vowel Perception Task. They heard English homophonic word pairs containing a target sound spelled with one or two letters and indicated whether the two words contained the same sounds or not. For half of the listeners a picture was used to activate target words (Auditory-Visual Input group). Bilinguals in this group perceived different sounds in homophonic pairs. Experiment 2 tested whether naturalistic exposure reduces orthographic effects on speech perception by comparing learners, sequential bilinguals and English controls (all n = 30) with Auditory-Visual Input. Orthographic form (‘spelling’) affected consonant perception in both of the second language listener groups. Learners were less affected than bilinguals. Analyses indicated that this was because of the learners’ high proficiency. It appears that ItalianL1 speakers of EnglishL2 make a long-short contrast for consonants — unattested in English — and illusorily perceive it in spoken English homophonous words.