Outline of a Model of L2 Phonological Learning

Outline of a Model of L2 Phonological Learning

by James Emil Flege

The following hypotheses can serve as the basis for a model of the acquisition of L2 phonology.

1. Every human being is born with “phonetic learning ability”, i.e., the ability to learn to identify the phonetic categories of an input language(s) and to produce speech with acoustic properties closely conforming to the phonetic characteristics of those categories.

2. L1 phonology “develops” in the following ways:

(a)   Phonetic categories are established before phonemic categories.  The number of phonemic categories will be smaller than the phonetic categories use to implement them when phonemes are produced with clearly identifiable allophones that are not phonetically conditioned; The number of phonemic categories is determined only after a sufficiently large lexicon has been established.

(b)  The perceptual representations for phonetic categories are elaborated until they conform closely to those of mature speakers in the surrounding community;

(c)   The basic motor plans specified in each phonetic category, and the sensorimotor realization rules used to translate the phonetic categories into articulatory gestures, are aligned with perceptual representations so that phonemes are produced in a language-appropriate manner;

(d)  To varying degrees, all of the above are dependent on the quantity and quality of the input received by the learner.

3. Phonetic learning ability remains intact through the lifespan.  Speech learning in L2 differs from L1 speech learning, however, because:

(a)   The phonetic system is gradually optimized for the encoding and decoding of the sounds in L1 (as outlined in 2);

(b)  The phonetic system becomes resistant to the addition of new phonetic categories, which is partly a result of the optimization process.

4. Whether L2 learners identify phones of L2 as “similar” or “new” has important consequences:

(a)   If similar, learners will substitute sounds from their L1 repertoire;

(b)  If new, learners will eventually produce them independently of sounds in the L1 repertoire. 

5. An L1 “accent” in the L2 may result from:

(a)   Immature attempts at a new sound, which may lead to “developmental processes” resembling those of children learning L1;

(b)  Equivalence classification of similar sounds, which may lead to transfer errors in production;

(c)   Incorrect lexical representations (e.g., /lak/ for rock);

(d)  Correct central representations but immature realization rules;

(e)   Some combination of (a)-(d).

6. Age of learning will determine how similar but new sounds in an L2 are treated:

(a)   Similar sounds are identified increasingly as being inside the phonetic repertoire and less often as being inside the phonetic repertoire and less often as being outside the repertoire as age of Learning (AOL) increases.

(b)  For individuals who begin learning an L2 before the age of about 5-7 years, additional phonetic categories are established for similar L2 sounds.  The corresponding L1 and L2 sounds will be implemented using different phonetic categories and phonetic realization rules.  The production of similar L1 and L2 sounds will be authentic.

(c)   For individuals who being learning an L2 after about the age of 5-7 years, additional phonetic categories will not be established.  The corresponding L1 and L2 sounds will mutually influence one another because they are implemented using the same phonetic category.  Differences in production may result from the application of different realization rules, but the L2 sounds will not be produced authentically.

(d)  Given sufficient L2 input, L2 learners remain able, even as adults, to establish additional phonetic categories for new L2 sounds.  Many of those who do so will go on to produce new L2 sounds authentically.  Attitudes and motivation, as well as psychological factors may play a role in defining phonetic input in these instances, and so may have an impact on how well new L2 sounds are produced.