Comparing Different Algorithms of an Automatized Computer-Scoring Method for Spelling (Poster 13)

• In Event: AERA Poster Session Six
  In Poster Session: Literacy Poster Session Two

Abstract

The present study compared the use of different algorithms of an automatized computer-scoring method for spelling. English-speaking students were assessed at grade 2 on their spelling, word reading, vocabulary, and cognitive skills, and followed up to grade 4. Students’ spelling was first scored as correct and incorrect and then scored using edit distance which measures how distant a spelling is from the target word. Edit distance scoring was done with three different algorithms: Damerau-Levenshtein distance (number of edits), cosine similarity (letter frequency), and editex (phonological plausibility). Results indicated the desirability of edit distance scores when the spelling scores were compared for their predictability of grade 4 spelling and word reading while controlling for grade 2 literacy and cognitive skills.

Authors

• Youngsun Moon, University of California - Irvine
  Presenting Author

• Young-Suk Kim, University of California - Irvine
  Non-Presenting Author