dliflc student profiles - ut liberal arts · • correlations of dlab and outcome data –...
Post on 16-Aug-2020
1 Views
Preview:
TRANSCRIPT
DLIFLC STUDENT PROFILES
Dr. John Lett, Dean* Research and Analysis
Defense Language Institute Foreign Language Center
*The views expressed are those of the speaker and do not necessarily represent those of the DLIFLC, the Department of the Army, or the Department of Defense.
Outline
• Demographics of DLIFLC students • Selection and assignment to language
study at DLIFLC • Role of motivation and language choice in
achieving advanced language learning outcomes.
Selection for Military Language Training (IET)
Recruiting
ASVAB
DLAB
BASIC TRAINING
DLIFLC
Student Demographics
Age
Gender
Education
Service
Years in Service
Paygrade
Aptitude for Language Learning
DLIFLC Learning Environment
• Intensive, small classes • Six or more hours daily • 26 to 63 weeks • Proficiency-oriented approach • Graduation requirement:
L2, R2, S1+ on ILR Scale*
*http://govtilr.org
Description of DLAB • Multiple choice, ~90 minutes • Range 12-164, mean 100, SD 15 (~92, SD
~21 for 143K cases; current DLI means ~112-114)
• Four parts – Bio data – Spoken stress – Deductive rule application – Inductive pattern application
RESEARCH BASE
• Carroll’s four components – phonetic coding ability – grammatical sensitivity – rote learning ability – inductive language learning ability
• Other FL aptitude tests – MLAT (Carroll-Sapon) – PLAB (Pimsleur)
How well does it work?
Very well, thank you – Language Skill Change Project, mid-1980s
– Army Research Institute, 2000-2001
– DMDC study, 2007
– DLIFLC program data, early 1980s to present
LSCP* • 1903 US Army students • Korean, Russian, German, Spanish • Eleven blocks of predictor variables • Forced order-of-entry MRA of L, R, S,
course completion • DLAB added variance, especially for
Russian and Korean *Lett, J. A., and O'Mara, F. E., "Predictors of Success in an Intensive Foreign Language Learning
Context: Correlates of Language Learning at the Defense Language Institute Foreign Language Center," pp 222-260 in Parry, T., and Stansfield, C. W., eds., Language Aptitude
Reconsidered, published for CAL/ERIC by Prentice Hall Regents, Englewood Cliffs, NJ, 1990.
ARI Attrition Study
• Tracked 237 beginning students for a year
• From ARI PI’s outbrief:
– As in past, DLAB is best predictor of attrition DLAB 98 or less = 32% attrition DLAB 99 to 116 = 20% attrition DLAB 117 or more = 10% attrition
– Of DLAB scores above 117, 98% had prior language training Prior language training = 17% attrition No prior language training = 35% attrition
DMDC Study, 2007
• Examined AFQT and DLAB as predictors of language learning success at DLIFLC
• Employed three samples, using Markov chain Monte Carlo procedures and Bayesian modeling
• Findings: AFQT < DLAB < Multiple Hurdle
DLIFLC Program Data
• Correlations of DLAB and outcome data – Consistently significant – Restriction of range affects magnitude
• Eight-year data set: Ss meeting the DLAB minimums have… – Lower attrition rates – Greater success rates (2/2/1+)
• Caveat: High aptitude is a predictor—not a guarantor!
Success Rates by DLAB (FY02 example)
Motivation, Choice, and Outcomes
• Does it matter whether you get a language you wanted?
• How important is motivation to ultimate success at DLIFLC?
Who Studies What?
• Constraints – Minimum DLAB requirements (95, 100, 105,
110) – Needs of the Service
• Timing – Pre-arrival – Post-arrival
Upon Arrival in Assigned Language Department
Does It Matter?
• Study 1: FY90 – FY94 graduates in 8 languages
• Study 2: FY01-FY09 graduates in 9 languages
• Six languages in common: – Arabic, Chinese (M), Korean, Farsi, Russian,
Spanish
Methodology • Compare “choice” responses to language
learning outcomes – Three variants of coding choice responses:
• All five options • “5” vs “1-4” • “5-4” vs “1-3”
– Outcome measures: Listening, Reading, Speaking, and Success (2/2/1, 2/2/1+)
– Partial correlation and chi square analysis – Visual inspection
Findings
• Distribution of Responses
• Relationships of Responses to Outcomes
Nine-Year Responses No reply
Comparison of Responses
Relationships with Outcomes
• Earlier Study – Most correlations non-significant or too small to
matter – Most chi squares non-significant – Much variation across languages and skills
• Current Study – More significant chi squares – Much variation across languages and years
• Caveat: What does “choice” item measure?
Another Look at Motivation
• Language Skill Change Project – 1903 students, 4 languages – Multiple Regression, 9 blocks of predictors
• Forced order-of-entry • Attitude-Motivation measures in blocks 6 and 7
• Motivation at start less predictive than motivation during
• Cognitive variables most predictive
The Bottom Line
Comments or Questions?
John.Lett@us.army.mil
top related