Lexical Quality of ESL Learners Effects of Focused

Lexical Quality of ESL Learners: Effects of Focused Training on Encoding Susan Dunlap, Benjamin Friedline, Alan Juffs, & Charles A. Perfetti University of Pittsburgh Jeanine Sun Washington University in St. Louis

Background • ESL encoding task (of RSAs) • Arab L 1 seem to make more spelling errors than Korean, Chinese, and Spanish L 1 • Differences cannot necessarily be accounted for by L 1 writing system, L 1 orthographic depth, L 2 vocabulary knowledge, or L 2 fluency

Previous Research • Arab L 1 have more problems with prelexical word identification; Japanese L 1 have more problems with online word integration (Fender, 2003)

Previous Research • Reading skill better than L 1 as a predictor of L 2 spelling accuracy in school-aged children (Wade-Woolley & Siegel, 1997)

Theoretical Framework • Lexical Quality Hypothesis – (Perfetti & Hart, 2001) in L 1 – orthography, phonology, meaning – plus don’t forget: syntax and morphology • L 1 affects L 2 learning of grammar, spelling, vocabulary, etc. – (Mac. Whinney, 2005)

Connection to PSLC Framework • Robust Learning – Retention (of trained words) – Transfer (to new words) – Accelerated future learning (faster decrease in error rates across ESL years) • Assistance dilemma • Explicit vs. implicit instruction

Hypotheses/Predictions • Intervention with focused encoding and meaning-based encoding task will increase quality of lexical representations – Retention • improved lexical quality (of trained words) – Transfer • improved lexical quality (of new/untrained words) – Accelerated future learning • faster decrease in error rates (steeper slope)

Method • Two-phase approach – Phase 1: Knowledge Component Analysis – Phase 2: Focused Intervention

Method • Phase 1 – Knowledge Component Analysis – in-depth coding of RSA transcription data – aka data mining

Coding • Correct – AWL K 1 -5 – acceptable (e. g. , accumulation, techniques) (e. g. , blog, otolaryngology, falafel) • Typing (form) – capitalization – punctuation – spacing • Errors – encoding errors (e. g. , english) (e. g. , couldnt) (e. g. , myfriend)

Error Types • Consonant – Missing – Extra – Substitution • Vowel – Missing – Extra – Substitution • Multiple C/V errors • Transpositions • Lexical/morphological – Plural, tense, affixes • Garble conect (spa 4) fittness (kor 3) afternoom (kor 4) tuch (chi 4) aabout (ara 4) becose (kor 3) voleyboll (spa 3) afetr (ara 3), becuase (kor 5) truthable (kor 4); laught (tai 3) cabegle (chi 4); thr (ara 4)

Preliminary Findings

Preliminary Findings

Preliminary Findings

Summary of Preliminary Findings • For all L 1 groups, errors decrease from Level 3 to Level 5 • Arab L 1 group makes more errors compared to other L 1 groups, this difference persists through Level 5 • Arab L 1 seem to be attempting more “advanced” words (fewer AWL 1 words) • Vowel errors most prevalent for Arab L 1 • Consonant errors most prevalent for Spanish L 1

Method • Phase 2 – Intervention – Fall 2008 – In vivo ESL Learn. Lab – Designed to focus attention to form-meaning mappings

Implementation • Participants – Pilot in Fall 2008 (Level 5 students) – Data collection in Spring 2009, weeks 1 -15 – ESL 3, 4, and 5 writing classes • Exercises – Required but not graded – Done in language lab (CL G-17) – Overseen by researcher on site for weekly scheduled lab times • Programmed in Revolution (or Flash? ) • Separate from REAP-based vocabulary study

Predicted Results • L 1 x Level x Focus (whole word/sublexical) – Retention • improved lexical quality (of trained words) – Transfer • improved lexical quality (of new/untrained words) – Accelerated future learning • faster decrease in error rates (steeper slope)

Acknowledgments • Sally J. Andrews, Michael Nugent, Claire Bradin Siskin • PSLC ESL Learn. Lab, funded by NSF award number SBE-0354420