A tDCS Study of Implicit Learning of Foreign Cognate and Non-Cognate Words Joshua S. Payne*, Paul Mullins & Marie-Josèphe Tainturier Centre for Research on Bilingualism and School of Psychology, Bangor University
Hypotheses
Background • Left STG/SMG is involved in the acquisition and maintenance of form-meaning connections.
• Cognates should be learned and retained more easily
• Anodal tDCS over left STG/SMG regions improves learning of pseudoword vocabularies.
• Greater learning gains during active versus sham stimulation
• First study to explore the effect of tDCS on learning foreign language vocabulary.
• Better performance on backward translation post-tDCS, up to a week later
• Few direct manipulations of psycholinguistic factors, such as cognate status.
• Differential effects of tDCS for cognates and non-cognates
• tDCS effects seem to be stronger for harder task or in less skilled participants
• Differential tDCS effects as a function of learning ability or phonological memory.
Methods Participants • 32 healthy young adults • Monolingual English speakers • Right-handed • No previous knowledge of Dutch
tDCS Parameters • Single-blind, 1mA tDCS • Active: 20mins (15s ramp on/off) • Sham: 30s (15s ramp on/off) • Anode (35cm2) over CP5 (left STG/SMG) • Cathode (35cm2) over right supraorbital
A: Protocol Day 1
Learning Task A 5 Blocks
Translation A
Translation A
Learning Task B 5 Blocks
Follow-up (+1 Week)
Day 3
Day 2
Translation B
Translation B
Translation A
Background Testing
Translation B
Stimuli & Materials • Two 30-item Dutch vocabulary lists • 15 Cognates, 15 Non-cognates
• Foils: miniature phonological neighbourhoods • ½ similar (P+), ½ dissimilar (P-)
• Implicit learning task adapted from Flöel et al. (2008) • 10:1 Correct:Foil pair ratio by Block 5
• Backward translation immediately, 1-day and I-week after learning session COMETS2 head model and current density simulation.
Results tDCS Effects on Learning FL Vocabulary
tDCS effects as a function of Immediate Translation Performance
Implicit Vocabulary Learning Task
(n = 17)
(n = 15)
• Groups split around the mean of average translation performance, immediately after stimulation • Poorer translators learned more non cognates during active stimulation vs. sham. • Effects evident from block 3 onwards •
Translation x Stimulation x CognateStatus (β = 0.28, p = .005)
• Good translators show a tendency toward worse performance during active stimulation.
Backward Translation Task
tDCS effects as a function of Phonological Memory Ability (n = 14)
(n = 18)
• Groups split based on bimodal distribution around the mean phonological memory score. • Participants with lower phonological memory scores are able to learn more non cognates during active stimulation vs. sham. • Effects evident from block 3 onwards •
PhonoMem x Stimulation x CognateStatus (p < .003)
• Participants with good phonological memory show a tendency toward poorer performance during active tDCS.
• Large learning effects, and overall large cognate advantage during learning and translation. • Cognates are retained better at follow-up compared to non cognates. • No significant effects of stimulation in either task.
Discussion • Cross-language phonological similarity facilitates acquisition of FL vocabulary.
• However, our results reveal that tDCS does improve learning of non cognates for poorer learning and/or learners with weaker phonological memory.
• Cognates are retained for a longer period of time following implicit learning. • Overall, active tDCS did not affect performance compared to sham.
• Item and individual level responses to tDCS consistent with a number of recent studies. • Multiple tDCS sessions paired with learning may show more robust effects of tDCS.
Contact: Joshua Payne
E:
[email protected]
T: +44 (0)1248 382925