Document Type


Date of Degree

Spring 2015

Degree Name

PhD (Doctor of Philosophy)

Degree In

Speech Pathology and Audiology

First Advisor

Tomblin, J B

First Committee Member

McGregor, Karla

Second Committee Member

Owen Van Horne, Amanda

Third Committee Member

Samuelson, Larissa

Fourth Committee Member

Duff, Melissa


Purpose: This study investigates one possible reason for individual differences in vocabulary learning from written context. A Latent Semantic Analysis (LSA) model is used to motivate the prediction of a causal relationship between semantic knowledge for words in a text and the quality of their hypotheses about the semantics of novel words, an effect mediated by reading comprehension. The purpose of this study was to test this prediction behaviorally, using a within subject repeated measures design to control for other variables affecting semantic word learning.

Methods: Participants in 6th grades (n=23) were given training to increase semantic knowledge of words from one of two texts, counterbalanced across participants. After training, participants read untreated and treated texts, which contained six nonword forms. Measures were taken of reading comprehension (RC) and the quality of the readers' hypotheses about the semantics of the novel words (HSNW). Text difficulty and semantic informativeness of the texts about nonwords were controlled.

Results: All participants had increases in semantic knowledge of taught words after intervention. For the group as a whole, RC scores were significantly higher in the treated than untreated condition, but HSNW scores were not significantly higher in the treated than untreated condition. Reading comprehension ability was a significant moderator of the effect of treatment on HSNW. A subgroup of participants with lower scores on a standardized reading comprehension measure (n=6) had significantly higher HSNW and RC scores in the treated than untreated condition. Participants with higher standardized reading comprehension scores (n=17) showed no effect of treatment on either RC or HSNW. Difference scores for RC and difference scores for HSNW were strongly related, indicating that within subjects, there is a relationship between RC and HSNW.

Conclusions: The results indicate that for a subgroup of readers with weaker reading comprehension, intervention to enhance lexical semantic richness had a substantial and significant effect on both their reading comprehension and on the quality of hypotheses that they generated about the meanings of novel words. Neither effect was found for a subgroup of readers with stronger reading comprehension. Clinical and educational implications are discussed.

Public Abstract

Reading comprehension and vocabulary are important academic outcomes. Between individuals, vocabulary skill is closely related to reading comprehension skill. This study investigated a possible reason for this relationship, namely that rich knowledge of word meaning could lead to better reading comprehension, which could lead to better inferences about the meaning of new words. This was investigated with an intervention with sixth graders that increased their knowledge of the meanings of a set of words. Participants then read one text that contained many of the taught words, and one text that did not. The study looked at the effect of the intervention on participants’ reading comprehension, and their definitions of new words in the texts.

The intervention resulted in increases in knowledge of the taught words, and in comprehension of texts containing the taught words. For the group as a whole, it did not result in significant improvements in the quality of definitions that participants generated for new words embedded in the texts. However, the effect of the intervention depended on the initial skill level of the participants. For participants who had lower reading comprehension skills initially, the intervention to increase knowledge of the meaning of words had a large effect on both reading comprehension and definitions of the meanings of novel words. This was not true for those who had higher reading comprehension initially.

The research has implications for teaching and clinical practice, suggesting that direct vocabulary instruction can result in benefits to learning of untaught words for some readers.


publicabstract, Individual differences, Latent Semantic Analysis, Reading, Reading Comprehension, Vocabulary, Word Learning


x, 128 pages


Includes bibliographical references (pages 98-107).


Copyright 2015 Dawna Margaret Duff