Shea & Curtin 2010

Discovering the relationship between context and allophones in a second language: Evidence for distribution-based learning

Q:Are learners aware of the factors that condition distribution of allophones?

Experiment 1
Method: Yr 2 and Yr 3 Spanish LLs choose which syllable is stressed
Stimuli: CVCV 2-syllable words; C are [b] [d] [g] or approximates; V alternate stress
Purpose: Determine association of stress with stop-initial syllables
*Subjects have all had explicit instruction on the stop-approximate alternation

Results:
NES and Yr 2 ratio of stop+stress:approx+stress was around 1 (chance)
Yr 3 ratio ~2
NSS ratio ~2.75
-- More experience led to greater association of stress with the stop
*All groups show trochee bias, only NSS additionally show bias for the stop

Experiment 2
Method: same as Exp 1
Stimuli: same as Exp 1 but vowel stress was neutralized; duration of consonants was manipulated
*So any perception of stress is an illusion!
*Duration of consonants was non-significant

Results:
NES and Yr 2 ratio of stop+stress:approx+stress was around 1 (chance)
Yr 3 ratio ~1.9
NSS ratio ~2.4
All showed a bias toward stop-onsets

--More experience led to greater association of stress with stop

My reaction:

• The illusory stress evidence is fascinating
• That there was that strong of a difference between yrs 2 and 3 is surprising
• Would there be differences without explicit instruction?
• Lots of different statistical tests were run to account for all kinds of bias