Transcript 48x72 poster template - University of Florida

Lack of adjustment in L2 English?
Eunjin Chun, Martha Hinrichs, and Edith Kaan
Department of Linguistics, University of Florida
INTRODUCTION
RESULTS: Cumulative adjustment
RESULTS: Verb-based surprisal
• Native speakers show adjustment to recently used syntactic structures.
• This has been explained by implicit, error-based learning based on predictive
processing (Cheng, et al. 2006).
• The larger the prediction error (e.g., the infrequent the structure), the larger
the adjustment.
• If second-language processing involves error-based learning, we expect L2
speakers to adjust to the preceding structures, and more strongly to structures
that are infrequent to them.
• Both Native English and L2 group show cumulative adjustment:
- Increased production of PO (in neutral prime trials) after having encountered
more PO sentences
- Decreased likelihood of producing PO after having encountered more DO
structures
• Steeper slopes for infrequent structures (PO for English; DO for L2)
• Prime verb bias:
- No effect of prime verb bias (no verb-based surprisal, contra Jaeger & Snider,
2013, but see Peter et al., 2015, for absence of surprisal effects in adults)
• Target verb bias:
- Both native and L2 group show effects of target verb bias, regardless of
priming: fewer POs produced with verbs that are less biased towards a PO
- Responses in priming study pattern with responses obtained in an independent
norming study
QUESTIONS
• Do L2-learner show adaptation to recently used syntactic structures?
• Cumulative adjustment (over the course of the study)?
• Immediate priming effects?
• Verb-surprisal effects: stronger priming if the prime has a structure that is
unusual given the prime verb (Jaeger & Snider, 2013)?
METHODS
• Participants:
• Native group: 72 native speakers of American-English (age 18-52)
• L2 group: 62 Korean intermediate English learners, recruited in South
Korea (age 18-27; started learning English at age 6, mean 9.7)
• Task and Materials:
• Web-based written priming study: describe pictures by completing a
sentence fragment
• Prepositional object (PO), Double object (DO) or neutral primes (9/
condition)
• Verb bias information collected in an independent completion study
without priming with a different group of native English and Korean L2
English
• Analysis (logistic mixed-effects model):
• Dependent Variable: Likelihood of PO response (versus DO and
Transitive responses)
• As a function of the number previously completed PO (DO) structures as
prime or target, prime condition, language group
RESULTS: Immediate priming
• Korean L2 English: use more PO than DO, use more Transitive
• Native English: use more DO than PO, use hardly any Transitive
DISCUSION AND CONCLUSION
• Do L2-learners show adaptation to recently used syntactic structures?
• Cumulative adjustment? (over the course of the study)
- Yes
- More so for infrequent structures
- Suggest increased baseline activation of abstract structures
• Native English
- Priming for less frequent structure (PO):
- More PO produced after a PO
prime than after neutral
- Only for first few trials
 Expected under error-based learning
• Korean L2 English
- No significant effects of priming
- No modulation of priming by
cumulative persistence
- …even though DO is infrequent
 Not expected under error-based
learning
• Verb-surprisal effects?
- No, but Native English do not show surprisal effects, either
- L2-learners are sensitive to Target verb bias  have verb-specific
representations
• Immediate priming effects?
- No, not even for infrequent DO structure
- Could be due to underspecified representation of DO structure
- Korean L2 English learners mostly use transitives as non-PO alternative:
[V …NPtheme] instead of [V NPpatient NPtheme]
- If the prediction for a PO is not met (i.e. if input is [V NPrecipient…] rather
than expected PO [V NPtheme…]), it is not immediatelty clear that the nonPO structure needs to be boosted.
• Adaptation effects in L2 speakers can be accounted for by error-based learning.
Lack of immediate adjustment in L2 processing can be explained by differences
in representation and frequency of the structures involved.
CONTACT
printed by
www.postersession.com
[email protected];
[email protected]