Role of Semantic and Morphological Information in Word Recognition and Its Instruction

• In Event: Mapping Morphology and Its Multidimensionality, Assessment, Relations to Outcomes, and Relation to Covariates

Abstract

Purpose:
This paper addresses (Question 1) how much variance in word recognition is explained by semantic or morphological variables—to index the potential for semantics to influence word recognition skills and (Question 2) whether instruction focused on semantics or morphology shows evidence of increasing word recognition skills.

Theoretical Framework:
For more than 50 years, there has been disagreement about whether readers should learn to use context and picture clues for word recognition. Goodman (1967) argued readers will use as little letter-sound information—and as much semantic information—as possible to recognize the word. The consequence is that students should focus on any meaning clues provided by the text—including prior and subsequent context and images on the page. Although scientific data suggest that readers use as much information as possible (letters, bodies, etc.) to recognize words, the data—and the underlying theory—do not obviate the importance of semantic information surrounding an unfamiliar word and perhaps also sublexical semantic information (i.e., morphemes), as important for word recognition—and, by extension, instruction. The instructional question is important in the “science of reading,” a core issue in which is whether or how much foundational word recognition instruction should focus on context-, picture-, or morpheme-based strategies that will foster the use of semantic cues.

Method:
Both research questions were answered with a meta-analysis involving a database literature search (3,699 records), additional reference chasing (30 additional records). Question 1 was answered using the partial correlation between semantic and word recognition variables after partialling out the effect of phonological skills. Question 2 was answered by calculating Hedges’ g using posttest scores.

Results:
Question 1 indicated that among the 37 included studies, the mean weighted partial correlation between word recognition and semantics was .27 [.19-.35] and between morphology and word recognition was .43 [.32-.53]. Among the 8 morphology-focused instructional studies for Question 2, the mean weighted effect size for morphology instruction was 0.15 [-0.03-0.34]—though several studies were conducted with middle school students. No studies of instruction were identified that focused on using semantic knowledge including context or picture clues to improve word recognition. Even a post hoc search specifically focused on context and picture strategies did not lead to identification of any studies.

Significance:
Semantic and morphological knowledge are associated with word recognition, but semantics explains relatively little variance after partialling out phonological variables. There may be a somewhat low limit on the potential for semantics to influence word recognition; morphology may be more powerful given that the sublexical units overlap with both phonology and semantics. This aligns with the possibility that morphological instruction might improve word recognition, particularly in elementary-age readers. The lack of empirical data on semantic strategies was disappointing considering the frequent focus on semantic strategies in beginning reading programs. The results indicate that experimental research is essential to help the field resolve this important ongoing issue; additional studies will add much-needed science to the science of reading debate.

Authors

• Devin Kearns, University of Connecticut - Storrs

• Melodee Anne Walker, The University of Texas at Austin

• Shannon Kelley, University of Connecticut - Storrs

• Cheryl Lyon