Kazuya Saito, from Birbeck Uni in London, has presented the results of his research on late L2 acquisition. "Late" acquirers would be those individuals who start learning a second language after puberty, here defined as people aged 16 onwards. His study involved Japanese speakers in their late teens and adults who had settled in Canada.
Find my (informal) report on the event below, with a few of my own comments, of course. (The usual disclaimer: all misinterpretations or misrepresentations of the info presented are my own)
***The talk began by reviewing two well-known models of L1 and L2 acquisition (or actually, "learning"? That is a whole separate debate....): the Speech Learning Model (SLM) by Flege (1993, 003, 2009), and the Critical Period Hypothesis (for this, Saito quotes Abrahamsson 2012, DeKeyser 2013, but the CPH dates from much earlier). Saito claims that these theories trigger different predictions as to what could happen to late learners.
The SPM would predict that enough exposure ("experience effects") to L2 will help learners to invoke those speech learning abilities that we applied for our L1, so age will have an effect on their ultimate level attainment, and so will the length of residence in a foreign country. The CPH, on the other hand, predicts that near nativelikeness is not attainable after puberty, which would affect late learners, and it also establishes that the skills employed for the learning of a second language involve general cognition -that is, explicit and intentional processes - rather than the explicit and incidental language-specific processes. (This latter point, I think, is very, very important when it comes to favouring explicit pronunciation instruction. However, the role of explicit pronunciation instruction was somehow argued against -or downplayed- in this presentation). In other words, for the CPH, the effects of the length of residence will also be limited.
This lit review finished with references to conflicting results in studies related to the effect of length of residence and age (mental note: go back to Linda Grant's (2014) edited volume for examples!). There was also a reference to something that is highlighted in other studies I have read, the role of motivation and aptitude as well in determining the levels of ultimate attainment. What is more, the presenter made some slight criticism to the methods employed in some of the studies, as they focused on native speaker evaluation of accents globally. Thus, Saito's "niche" lies in the focus on one specific pronunciation feature: the Japanese flap [ɾ] vs English /r/. (Two comments: I have used /r/ for the English version because it was not very clear to me whether the target was an actual retroflex approximant [ɻ], or the alveolar [ɹ] one. And for the Japanese sound, the alveolar tap symbol was used, but in the speaker's description, I often wondered if it was not the alveolar lateral flap [ɺ] that was referred to. The Handbook of the IPA says that for Japanese it's [ɽ], actually, so I should have asked! ).
The difficulties that Japanese speakers face when it comes to English /r/ are threefold: two of these are related to adjusting cues already present in Japanese: the retraction of the tongue body -shown as the lowering of F2 values, acoustically speaking -(which is why I guess the target /r/ is a retroflex, perhaps) and the prolongation of length/duration; whereas the new feature for these learners would be the presence of labial, alveolar and pharyngeal constructions, manifested acoustically in the lowering of F3.
The study was aimed at discovering some patterns in terms of the effect of AOA (age of acquisition) and LOR (length of residence), and it involved a number of Japanese people who had moved to Canada after they had turned 16, and most of them had only had instruction in English at school -using a Grammar Translation method, according to the presenter-. They have all been described as being "highly motivated" to attain a high level of English because of their need to communicate. The experiment consisted of testing the subjects' production of English /r/ in a picture-description spontaneous production task (adapted from Munro and Mann, 2007). In all the pictures there were target words with /r/, and in order to make sure participants were tested in a more spontaneous situation, the first 3 descriptions were presented as being "for practice", and the other four were used for the test.
The results of the study revealed the following:
- In terms of tongue retraction, most participants improved over their first six months of residence, and after that period, in this respect, their performance was deemed native-like.
- As for duration, the correlations were significant after 12 months of residence.
- When it came to the development of the new parameters (F3), levels of attainment were diverse and late acquirers were definitely at a disadvantage (only those moving to Canada before the age of 20 achieved better results in this study), but definitive changes were seen after 10 years of residence (!).
Some implications for teaching were mentioned later. (I have to admit that this is perhaps where I would disagree the most, but it is true that I have worked in non-immersion contexts, and things are definitely different there). Based on these results, Saito contended that an effective tool is the use of contextualised instruction over explicit instruction of the target sound. Some of the activities mentioned included the use of prompted discussion and role plays with target words that were corrected on the spot. Some included /l,r/ minimal pairs, others did not. The testing of these activities after training students for 4 hours in two weeks rendered an improvement in students' /r/ sounds from 60% to 75%. (Plus, the "shock effect" of constant correction cannot, of course, be underrated, if these are highly motivated students!) In my view, explicit pronunciation instruction (because of the employment of these cognitive procedures mentioned earlier) is necessary to help create these new articulatory habits. I do, however, strongly make a case for contextualised and communicative pronunciation work, but not as the entry point to learn the sounds, but to make explicit instruction somehow transferrable to more spontaneous contexts (and this is, after all, the end result that Saito was seeking).
There were some very brief, but interesting questions. The issue of U-shaped learning and plateaus that is described in some psycholinguistic theories (if I remember correctly, Major's Ontogeny model, for example?) was brought up, inviting in a way the exploration of a more diachronic study of the same subjects. Another issue mentioned is whether a similar effect would be reached for the training of VOT as a low functional load feature, which was considered non-distinctive (well, I may have disagree with that, I believe that both VOT and vowel length can be distinctive in native-speaker's ears, no matter their allophonic status!).
All in all, it was a very interesting and thought-provoking talk with many implications for pronunciation teaching, both in immersion and other ESOL contexts. And a confession: at first, I was doubtful about attending a talk about Japanese, but I have to admit this whole uni experience has really widened my horizons to all the beauty and human and cultural wealth there is in this world.