Previously, we found that in terms of standard durational ratios, even the beginners appeared to be rather successful in distinguishing between L2 singleton and geminate sounds in their production. However, it was also possible that they were merely substituting real geminates with the coping strategy of an unreleased coda + homorganic initial consonant sequence. To this end, we analysed the mean intensity data of nonwords sassa, sesse, and sesso, to investigate whether the learners were producing a coda /t/ + initial /s/ sequence, presumably transferred from L1.
The results showed that the beginner group were indeed using this ‘cat-tail’ strategy, whereas the advanced learners and the native speakers were producing a genuine geminated /s/. We conclude that although derived geminates in L1 do not help beginners acquire L2 underlying geminates, it can be learned after enough exposure.