Introduction This study examined how long-term immersion in a second language (L2) affects the acquisition and maintenance of long vowels in bilinguals whose first language (L1) is Arabic or English. Focusing on Arabic /aː/, /iː/, /uː/ and English /ɑː/, /iː/, /uː/, the study investigated whether highly proficient late bilinguals produce nativelike vowels in both their L1 and L2, whether sound discrimination aptitude predicts vowel production accuracy, and to what extent L2 acquisition and L1 attrition are interrelated. Methods Vowel productions by highly proficient Arabic-English and English-Arabic late bilinguals were compared with those of monolingual control speakers. Productions were analyzed in terms of vowel height (F1) and vowel frontness (F2). Linear mixed-effects models and correlation analyses were used to assess group differences and relationships between L1 and L2 nativelikeness, as well as the contribution of sound discrimination aptitude. Results Results revealed vowel-specific and asymmetric patterns. Both bilingual groups produced nativelike /aː/–/ɑː/, reflecting successful L2 acquisition and stable L1 maintenance. In contrast, /iː/ and /uː/ showed cross-linguistic influences: Arabic /iː/ exhibited evidence of L1 attrition, while /uː/ displayed both non-native L2 realizations and modified L1 productions. A positive association between L2 and L1 nativelikeness emerged for /uː/, suggesting parallel proficiency across languages. Sound discrimination aptitude had little effect on vowel production accuracy. Discussion Overall, the findings support dynamic models of bilingual speech production, such as the revised Speech Learning Model, by demonstrating that L1 and L2 vowel categories remain interdependent. The results highlight that both individual and contextual factors contribute to the phonetic stability and modification of bilingual speech, and that L2 acquisition and L1 maintenance can proceed in parallel for specific phonetic categories.
Alharbi et al. (Wed,) studied this question.