The BAF Project has been running for over twenty years and has provided a most valuable framework for studies relating to rate of language-learning and optimal age for language acquisition in school contexts.

Amount of L2 input at an early age combined with length of exposure leads to successful L2 learning.

The study compared the performance of bilingual Catalan-Spanish learners who began a first and / or second foreign language (FL) at ages 4, 8, 11, 14, and 18+ years (N = approx. 2,000). Participants were drawn from a range of state schools (N = 30) and varying socio-economic areas. Analyses included comparison of longitudinal and cross-sectional data. For a detailed description of the project, see: Muñoz, 2006.

Composition of groups
> Children learning FL English who received approximately 750+ hours of instruction, some over seven years at a more intensive rate (3 hours per week from age 11), some over ten years at a less intensive rate (2 and 2.5 hours per week from age 8). 
> Children beginning FL English at age 14 (N = 51) having previously learned either FL French, and children learning FL French and FL English concurrently. 
> Adults (N = 135) learning FL English at age 18+ (with previous FL French experience in school)
> The pre-school group is not included here.

Groups were compared on a range of L2 achievement measures at three time intervals: after 200 hours, 416 hours and 726 hours of instruction.

Instruments 
The same measures were used for all groups. Although not age-appropriate, they were to some extent graded in difficulty, and exhibited satisfactory reliability (Cronbach’s alpha). Test measures comprised: dictation in Spanish, Catalan, and English; cloze tests in three languages; listening; writing; oral narrative, and interview; grammar; role-play; phonetic imitation, and discrimination.

Comparison of longitudinal data: learners beginning early (age 8) and late (age 11)
Measures: cloze, listening, and dictation.

Starting age 
Both time and group were statistically significant, which indicates substantial improvement in learning over time, and differences in starting age. The differences were highly significant in the cloze test and the dictation test after 200 hours and 416 hours of instruction in favour of the older starters, and likewise for the listening comprehension, test, but only after 416 hours of instruction, which suggests that in the beginner stages of listening, early and late starters performed equally well.

Rate of acquisition
The interaction for group and time was also significant. The late starters (age 11) displayed a much faster rate of improvement on all tests, although for listening comprehension the rate of learning at 200 hours was similar for both groups, only increasing for the late starters after this time.

Comparison of cross-sectional data by initial age of learning and length of instruction
Participants were distributed in groups according to initial age of learning and accumulated hours of instruction at the time of testing. Older-starting learners obtained higher scores at 200, 416 and 726 hours of instruction, with significant advantages across most tests. There was, however, a non-significant difference between 8- and 11-year-old starters after 200 hours on the cloze test and listening comprehension test, and only marginal significance on the dictation test. Likewise, there was no significant difference between 14-year-old starters and 18+ starters after 200 hours of instruction on the listening  comprehension. In sum, significantly higher scores related to older starting age, and consequently also to increased biological age at testing.

Do younger foreign language learners in instructed contexts eventually surpass older learners? Although the answer must be ‘no’, there are finer points to consider. Older learners, for example, obtained greater scores although these differences did not diminish homogeneously for all of the tests.

There were sometimes marginal differences between groups after 200 hours of instruction After 416 hours of instruction older children statistically outperformed the younger children on nearly all measures. After 726 hours the differences in both phonetic discrimination and fluency on some measures were no longer significant, and yet the general trend whereby older children outperform younger ones was still evident, and continued up to and beyond 700+ hours of instruction. There were, however, no significant group differences in listening comprehension after 200 hours, only after 416 hours.

Could there be different age effects for different language skills or indeed for any of the subcomponents? Comparisons of different test results taken at different times does indeed suggest that age effects are not uniform across measures of language abilities: “differences between older starters and younger starters vary in the different tests”, and “distinct patterns of development characterise the acquisition of different language subcomponents” (2006:30).

The largest inter-group difference was in the cloze test, the smallest in the test of listening comprehension (early stages), and in the reception measure on the oral interview, where beginners aged 8 catch up with beginners aged 11 after 726 hours of instruction. Phonetic discrimination became non-significant after 726 hours, and some differences in written fluency disappeared.

In sum 
> late starters outperform younger starters after a similar quantity of hours of instruction;
> different rates of acquisition: younger learners slower, with “speeded progression between the ages of 11 and 13, much faster than that between ages 14 and 16″ (2006:31);
> adolescent and adult beginners rapid initial rate of learning after 200 hours;
> 11-year-old beginners made most progress between 200 and 416 hours and 8-year-old beginners made most rapid learning between 416 and 726 hours with both showing an increased learning rate at age 12, which may be due to an increase in cognitive development.

“[F]indings suggest that second language learning success in a foreign language context may be as much a function of exposure as of age. Exposure needs to be intense and to provide an adequate model” (2006:34).

In school contexts where opportunities for implicit learning and practice are minimal, older learners may be quicker to acquire another language.

“[W]hen younger learners attain a state of cognitive development that is similar to that of the older learners with whom they are being compared, and are given the same conditions of time and exposure (and instruction), the differences should disappear. Initial age of learning seems more relevant for skills that can be acquired implicitly whereas age at learning can be seen as a factor explaining the rate of learning of most skills” (2006:34).