“…This suggests that differences in home language use across the two bilingual groups did not affect their vowel realisations in Welsh. To establish which Welsh vowel categories were acoustically closest to which English ones, a LINEAR DISCRIMINANT ANALYSIS was conducted, in line with previous studies (e.g., Williams & Escudero, 2014a -heed, hid-hid, hêd-herd, hed-head, hadhad, hâd-hard, hôd-hoard, hod-hod, hŵd-who'd, hwd-hood, hyd-hud) and language group (monolingual English or Welsh-dominant bilingual) and the random factor was speaker. Random slopes were not entered because vowel pair is not a factor repeated across all participants in the model as the tokens come from two experiments with different within-subjects items.…”