A Rose by Any Other Cognomen: Unpacking Predictors Multiword-Expression Knowledge With a Focus on Frequency

• In Event: Word Exposure and Vocabulary in Action: Exploring Frequency, Lexical Pragmatics, and Their Relationships

Abstract

Purpose
This study examines results from a novel measure of multiword expressions. We examine how target-word features predict item difficulty on this task relative to a synonym task assessing the same words. Our results indicate relationships between performance in both measures; especially word frequency. Yet, there are important differences that affect how we think about developing receptive and productive word knowledge.
Introduction
Traditional vocabulary assessments do not align with what we know about vocabulary knowledge. Varying assessments can help us better identify students' lexicons at different depths. While a student might not be able to identify a synonym for "controversy," they might identify the phrase "to spark controversy," which captures word knowledge missed by synonym identification; possibly incidental word learning from repeated exposure to expressions (Webb et al., 2013).
Simultaneously, little is known about what makes learning different academic words difficult. It is well-established that frequent words are more likely known, and complex words can be more challenging to decode. Yet, minimal research discusses how different target-word features influence ways of knowing a word. This study highlights the intersection between assessment and impact of word features on the probability of answering multiple-choice questions correctly. In doing so, we can further our understanding of how students learn words and how we can focus on different aspects of word features and knowledge in instructional practices.
Methods
We examined data from students in 6th-8th grade across performance on multiple-choice synonym versus multiword expression items. All students received 48 synonym items but only twelve multiword expression items.
We used a mixed binomial logistic regression to predict item correctness and model both person characteristics (e.g., grade level, reading comprehension, language status) and item characteristics (item type and word features). We included five latent factor word features about the target-word: Frequency (across corpora), Complexity (orthographic/phonologic), Proximity (neighborhood density), Polysemy (senses/meanings), and Diversity (context/semantic). Scores for each target-word are estimated using exploratory factor analyses across 22 word measures on a collection of over 1,000 academic words (Authors, under review). We allowed word features to interact with item type plus main effects.
Results
There was no overall difference in performance on each item. Regardless of item type, we found that more frequent, polysemous, and diverse words were easier (beta=.45, .49, and .89, respectively). However, the Frequency effect was significantly smaller for multiword expressions (beta=-0.29). While Complexity showed no significant main effect, complex words were significantly more difficult on multiword expression items (beta=-0.14). More polysemous and diverse words were also more difficult in the multiword expression—a complete reversal of the main effect (beta=-0.98).
Significance
Our findings support previous research that there are multiple ways to know a word and that some word features may make learning various aspects of a word diversely challenging. While identifying synonyms of high-frequency words is easier, that does not mean that other tasks are easier for frequent words. Our study highlights how teachers should be aware that there are important differences in how frequency effects support learning across students and across task types.

Authors

• Rebecca E Knoph, University of Oslo
  Presenting Author

• Joshua Fahey Lawrence, University of Oslo
  Presenting Author

• Åste Hagen, University of Oslo
  Non-Presenting Author

• Jin Kyoung Hwang, University of California - Irvine
  Non-Presenting Author

• Paulina Kulesz, University of Houston
  Non-Presenting Author

• David J. Francis, University of Houston
  Non-Presenting Author