Major, R. C. (2008). Transfer in second language phonology. Phonology and second language acquisition, 36, 63-94.
Major’s review touched upon some very interesting topics in the field of SLA.
Both the contrastive analysis theory and the markedness differential hypothesis seem to treat second language acquisition as a closed system. That is, the two theories seem to presuppose that L1 and L2 are the only two factors in play. Interlanguage is like a little magnet moving in between the two big metals (L1 and L2). “Markedness” and maybe “similarity” are like a magnetic field that pushes the little magnet toward either L1 or L2 until the magnet stops moving and the whole system reaches equilibrium. OT theory in its core shares the same presumption.
However, we know that the process of acquiring a second language is not a closed system. One’s age, prior experience, intelligence, education background, and so on all play a role in the acquisition of L2. That’s why I think we should pay attention to the issues raised on p.71, where Major briefly discussed issues other than the linguistic structures of L1 and L2.
Near the end of this review, Major emphasized that when compare the impact of “similarity” and “markedness” on L2 acquisition, one needs to make sure that all the other factors are strictly controlled for. I quite appreciate the intention, but is it possible to control for all the other factors, given the fact that we do not really know how many other factors there are. Since we cannot control for factors we do not yet know, the notion of controlling for all factors might lead to some practical and epistemological problems.
On p.65, Major mentioned the probabilistic framework of contrastive analysis in 1 line. I do wish he could have elaborated on this point a little bit further. The phenomenon of L1 default vowel (a probabilistic result)used as epenthetic vowel in L2, for example, could be a strong case in favor of L1 transfer. I think we cannot simply say if feature A exists in one’s L1, one will have no problem learning the same feature in his/her L2. We also need to know the probability for A to exist in certain environments. Mandarin speakers, whose native language has diphthong /eɪ/, often shorten /eɪ/ to /e/ in words like “train”. When investigate the probability for /eɪ/ to exist in a nasal environment in Mandarin, we get 0; while the probability for /e/ to exist in a nasal environment is relatively higher. The error of shortening /eɪ/ can therefore be argued as a prediction error of Mandarin speakers’ L1 model, rather than the preference for a less marked segment. I think a contrastive analysis on a segmental level is not sufficient enough; a probabilistic analysis taking into account “segmental environments” might shed new lights on the nature of L1 transfer.
Other issues such as the definition of markedness and the relationship between perception and production are not discussed in detail in this review. I am looking forward to learning your views regarding the aforementioned topics.
Zhiyan Gao, 9/10/2014