Few studies have investigated first language (L1) attrition and second language (L2) acquisition of segmental and suprasegmental aspects of speech. The current research includes three separate studies that examined L1 attrition and L2 acquisition of prosody of wh-words, vowels formants and the voice onset time (VOT) of voiceless stops in late consecutive Arabic-English (A-E) and English-Arabic (E-A) bilinguals. All three studies explored whether late L2 learners’ language use and phonetic aptitude (i.e. talent) affect the prosody of
wh-words, the formants of shared vowels and the VOTs of voiceless plosives among advanced learners of English and Arabic. Sixty participants participated in the three studies: 15 E-A bilinguals, 15 A-E bilinguals, 15 monolingual English speakers and 15 monolingual Arabic speakers. Both bilingual groups had been living in the L2 environment for about 20 years and had completed a proficiency test, a phonetic aptitude test and a language background questionnaire.
In study I, the participants read brief dialogues of wh-question/answer pairs; the
bilinguals read them in Arabic and English. To capture differences in slope steepness and the amount of curvature, the pitch contour of the wh-question words was analysed using growthcurve analyses. The monolinguals’ results revealed a steep rise to a high target in Arabic whwords, but no such high target in English wh-words. The bilinguals’ results revealed that A-E and E-A bilinguals approximated the prosodic patterns of the L2. An asymmetrical pattern of
L1 attrition was found, showing attrition among the E-A bilinguals, but not among the A-E bilinguals. Additionally, language use, but not phonetic aptitude, modulated how closely bilingual participants approximate native patterns.
In study II, the participants read words containing different vowels that are shared across English and Arabic in isolation and in a carrier phrase. The bilinguals read the vowels in Arabic and English. The first three formants (F1, F2 and F3) were extracted at vowel midpoints using Praat and then normalised. The monolingual groups’ results revealed differences in some of the shared vowels in Arabic and English. The bilinguals’ results revealed L1 attrition among the A-E bilinguals and the E-A bilinguals. Language use and phonetic aptitude were not found to influence the vowel formants of A-E and E-A bilinguals in L1 attrition and L2 acquisition. In study III, the participants narrated different cartoons, three in English and two in Arabic, and the bilinguals narrated them in both languages. VOT was measured using Praat from the interval between the plosive release and the onset of voicing. The monolinguals’ results showed that the VOTs of voiceless plosives differed in Arabic and English. As in study I, the bilinguals’ results showed an asymmetrical pattern of L1 attrition, showing attrition among the E-A bilinguals, but not among the A-E bilinguals. As in study II, language use and phonetic aptitude were not found to influence the VOTs of A-E and E-A
bilinguals in L1 attrition and L2 acquisition. The results suggest that L1 use may prevent L1 phonetic attrition, but only at the suprasegmental level. Some of the segmental results further support the Speech Learning Model (SLM): some areas of pronunciation are more vulnerable to attrition than others and some areas are more likely to achieve native-likeness than others. The present study revealed
that some aspects of native-like L2 acquisition can occur alongside L1 attrition, which means that while these aspects are acquired well in the L2, they are susceptible to attrition in L1. However, this is not true for all aspects. Finally, the findings confirm that acquiring a language from birth is not sufficient to ensure L1 native-likeness in the production of bilingual speech.