Word Learning During Reading: Effects of Language Ability in School-Age Children

Abstract

Most vocabulary growth during the school-age years occurs incidentally. However, little is understood about the influence of language skills on word knowledge growth during reading. Using a pretest–posttest quasi-experimental design, we examined incidental word learning through reading, considering the presence/absence of supportive context and the role of language ability. Children with a range of language abilities (N = 32), aged 10 years, 6 months to 16 years, 5 months, were exposed three times to rare nouns and verbs within stories or in isolation. Small but significant knowledge gains were found for rare words encountered in context, but not for words in isolation. Language skill predicted overall word knowledge but not rate of word knowledge growth. Findings suggest children with low oral language ability are at a disadvantage in acquiring vocabulary through reading; however, the word learning process may be qualitatively similar for children with varying language skill levels.

Keywords

language assessment reading vocabulary literacy

Children acquire new vocabulary at a rapid rate throughout the school-age years. It has been estimated that children learn about 3,000 words per year (Nagy & Herman, 1987) and that by age 10, a typically developing child knows close to 40,000 words (Anglin, 1993). Moreover, the majority of vocabulary growth during these years occurs in a seemingly effortless fashion, without formal instruction (i.e., through incidental exposure). Thus, there is general agreement that children’s incidental encounters with words, especially through reading, comprise the major source of vocabulary growth during the school-age years (see Nagy & Herman, 1987, for a discussion).

A strong vocabulary is especially crucial for proficient reading comprehension. Decades’ worth of studies have revealed strong and reciprocal relations across the life span between vocabulary and reading ability; evidence shows that skill in reading comprehension fosters the development of vocabulary, whereas a strong vocabulary enhances reading comprehension (for a recent review, see Adlof & Perfetti, 2013). Biemiller (2006) emphasized the crucial importance of vocabulary to school success and pointed out that although neither fluent decoding nor adequate vocabulary is sufficient for reading comprehension, if either skill is deficient, comprehension—and learning—will suffer. Hirsch (2003) argued that 90% to 95% of words in a passage must already be known for readers to achieve adequate levels of comprehension. It is clear that to succeed in school, children must be able to acquire new vocabulary through written context. Despite its importance, however, the process of incidental word learning during reading is still poorly understood.

Word Learning During Reading

Growth in word knowledge through reading is a gradual and incremental process. In a meta-analysis of studies examining incidental word learning during reading, Swanborn and de Glopper (1999) found that, on average, readers’ knowledge of unfamiliar words grew by 15% after a single exposure. Students in high school gained more word knowledge than younger students, and students with higher reading ability learned more than those with lower ability. From a textual standpoint, written passages with a lower density of target words were associated with greater word knowledge growth than passages with a higher density. Studies that included measures of partial word knowledge also reported greater gains than studies with measures that were not sensitive to partial word knowledge. In a later study, Swanborn and de Glopper (2002) reported that the extent of word learning during reading may also differ according to both reading ability and expected reading purpose. Sixth graders of higher reading ability gained more word knowledge if asked to read passages for comprehension or to learn the topic than if asked to read the passages simply for fun. However, students of lower reading ability acquired more word knowledge when asked to read for fun.

Language Skill and Incidental Word Learning During Reading

The extent to which language ability influences incidental word learning during reading remains largely unknown, and only a few studies have explored the association. Ricketts, Bishop, Pimperton, and Nation (2011) examined orthographic and semantic aspects of novel word learning during a read-aloud task. They explored factors related to word knowledge growth in an unselected sample of 7- to 8-year-old children with widely varying skills in reading and oral expressive vocabulary. The sample included participants with typical reading and vocabulary skills as well as participants with reading and vocabulary skills substantially below the typical range. Whereas only target word decoding predicted orthographic word learning, decoding, expressive vocabulary, and text reading accuracy predicted semantic learning. However, the study did not assess whether language skills other than oral expressive vocabulary predicted either orthographic or semantic word knowledge growth.

Whereas little is known about the association between language skills and incidental word learning during reading in general, even less is known about the association between language skills and incidental word learning during silent reading. In one of the few studies to address the relation, Steele and Watkins (2010) examined lexical acquisition during silent reading by comparing novel word learning among 9- to 11-year-old children with language learning disability (LLD) with that of age-matched typically developing peers. The LLD group had all been diagnosed with language impairment and/or learning disability in reading comprehension. After silently reading passages containing target novel words, children were assessed for syntactic and semantic knowledge of the words. Children with LLD scored significantly lower than the typically developing group. In an extension of this study, Steele (2015) found lower levels of semantic learning during reading in a group of school-age children with LLD than in groups of age- and vocabulary-matched typically developing peers, who performed similarly. Likewise, a group-design study of syntactic–semantic word knowledge growth during silent reading among the same participants included in the present study found that students with higher language ability outperformed their peers with lower language ability (Wagovich, Hill, & Petroski, 2015).

Growth in Word Knowledge Over Time

Most research findings concerning growth in word knowledge over time are based on oral incidental learning studies with young children. Taken together, results suggest that children with language impairment may require a significantly greater number of exposures to unfamiliar words, relative to typically developing children, to achieve a comparable level of vocabulary growth. For example, Rice, Oetting, Marquis, Bode, and Pae (1994) found that although typically developing 5-year-olds made significant gains in learning unfamiliar nouns and verbs after three repetitions, a group of children with specific language impairment (SLI) required 10 exposures to perform similarly. Likewise, Nash and Donaldson (2005) reported that for both incidental and explicit teaching contexts combined, children with SLI scored lower after 12 exposures than their age-matched typically developing peers did after six.

As for growth in oral word knowledge—a process whereby initial fast mapping (i.e., gleaning partial knowledge of a word through an initial exposure) is followed by the development of a fuller understanding of the word (Carey, 1978)—children’s incidental word learning during reading is an incremental process that unfolds gradually over time, with small but detectable amounts of word knowledge growth upon each encounter with a new word (Nagy & Herman, 1987; Swanborn & de Glopper, 1999). The number of exposures to an unfamiliar word should, thus, influence the amount of syntactic and semantic word knowledge children gain. However, only a few studies have examined how repeated exposure to novel or unfamiliar words might influence syntactic or semantic aspects of children’s incidental word learning during reading, and the findings are conflicting (Ricketts, Bishop, & Nation, 2008; Steele, 2015; Steele & Watkins, 2010; Wagovich et al., 2015). Steele and Watkins reported that the number of presentations (either two or five) did not affect word learning performance for children with either LLD or typical skills. However, Ricketts et al. found that both poor comprehenders and typically developing children benefited from an additional exposure to experimental nonwords, and Wagovich et al. reported an increase in word knowledge over the course of three presentations for readers with both higher and lower language ability. In contrast, Steele compared school-age children with LLD with both age- and vocabulary-matched typically developing peers on semantic word learning during reading and found that for children with LLD, a higher rate of presentation (5 times rather than twice) did result in greater word knowledge growth. Although both age- and vocabulary-matched controls performed better than children with LLD overall, neither typically developing group benefited from greater exposure.

Decoding skill and language ability are both likely to play a significant role in word learning during reading; although each set of skills is necessary, neither is sufficient. For example, among children with hyperlexia, who may demonstrate word reading ability as early as age 2, decoding of words and pseudowords often greatly outpaces the development of vocabulary as well as both oral and reading comprehension (Cardoso-Martins & da Silva, 2010). It is clear that children with weaker reading comprehension skills benefit less from text encounters with words in building vocabulary, leading to a vocabulary gap that increases over time (i.e., the Matthew Effect, Stanovich, 1986; e.g., Cain & Oakhill, 2011). However, given the conflicting findings, it remains unclear as to how language ability, either within or below the typical range, affects the pace of word knowledge growth with repeated exposure to words through reading.

Noun and Verb Learning

Only a few studies have explored the effects of form class on syntactic or semantic aspects of word learning during silent reading in school-age children, and among those that have, the results are conflicting (Schwanenflugel, Stahl, & McFalls, 1997; Steele, 2015; Steele & Watkins, 2010; Wagovich et al., 2015). Schwanenflugel et al. found that typically developing fourth graders’ word learning performance was higher for nonnouns than for nouns. However, Steele and Watkins did not find significant form class differences for syntactic–semantic word knowledge growth among either typically developing or LLD groups. Likewise, Wagovich et al. reported similar levels of syntactic–semantic word knowledge growth for nouns and verbs among school-age students of both higher and lower language ability. In contrast, Steele found greater semantic learning for nouns among children with LLD as well as age- and vocabulary-matched typically developing controls. Much still remains to be understood about the factors contributing to incidental noun and verb learning during reading. For example, little is understood about how form class might affect patterns of word learning over time, or to what extent language skills influence those learning patterns. Even less is known about whether form class exerts differential effects on the separate syntactic and semantic aspects of word knowledge growth during reading.

Syntactic and Semantic Word Knowledge

The use of syntactic cues to assist in gleaning the meaning of a new word (syntactic bootstrapping; Gleitman & Gleitman, 1992) is an important means of word learning from context for typically developing children as young as 17 months of age (Brown, 1957; Gelman & Taylor, 1984). In contrast, children with language impairment perform relatively poorly when using syntactic information to infer word meaning (Eyer et al., 2002; Rice, Cleave, & Oetting, 2000). In a study of typically developing fourth graders, children’s level of syntactic understanding had varying effects on semantic learning for words in different grammatical classes (Marinellie & Kneile, 2012). However, little is known about patterns of growth over time during reading for syntactic aspects of word knowledge, relative to semantic aspects. Moreover, it is unknown whether language ability has a similar level of influence on both syntactic and semantic word knowledge growth.

The Current Study

Given the central importance of incidental word learning during reading to children’s overall school success, the current study was designed to examine patterns of growth in syntactic and semantic word knowledge over repeated exposure to nouns and verbs through silent reading, among students with a wide range of oral language skills. We examined the role of written context on readers’ word knowledge growth, comparing participants’ word knowledge under two separate conditions: when unfamiliar/rare words were presented within the context of supportive narrative passages and when such unfamiliar/rare words were encountered in isolation. The research questions were as follows:

Research Question 1: How does word knowledge growth differ when words are presented in context, compared with in isolation?

Research Question 2: How does language ability affect word learning of nouns versus verbs, measuring syntactic versus semantic word knowledge, and in the presence of context versus no context?

Method

Participants

Thirty-two children and adolescents (24 girls, eight boys) participated in this study. Participants were the same as those in Wagovich, Pak, and Miller (2012) and Wagovich et al. (2015). Participants ranged in age from 10 years, 6 months to 16 years, 5 months (M = 13.24 years, SD = 1.58 years) and were recruited with flyers and electronic announcements posted in mid-Missouri. According to information provided by parents, 78% of the children were of European American origin, 15.6% were African American, and 6.4% were of mixed or Other ancestry. According to parent report, all the children were monolingual English speakers, all had normal vision, either corrected or uncorrected, and none had a history of neurologic impairment, either developmental or acquired. To participate, children were required to pass a bilateral hearing screening and earn a standard score of at least 7 (test M = 10, SD = 3) on the Symbolic Relations subtest of the Detroit Test of Learning Aptitude–4th Edition (SR-DTLA-4; Hammill, 1998), a screening measure of nonverbal reasoning.

Parents provided information about any diagnoses their children might have received for speech or language disorders and for learning or reading disabilities. Seven parents (21.9%) indicated that their children had been diagnosed with a speech disorder. Four parents (12.5%) indicated that their children had been diagnosed with a learning disability, six (18.8%) with a reading disability, and one (3.1%) with a language disorder. A subset of children was reported to be currently receiving services: four (12.5%) for language disorder and five (15.6%) for reading disability.

Parents also provided information about maternal educational status. The majority of participants had mothers with education at the college level. Overall, 27 of the 32 mothers reported having some college (with 20 of those being college graduates). Two participants had mothers with education described as non–high school graduate.

Tests and Experimental Measures

Standardized tests

The Clinical Evaluation of Language Fundamentals–4th Edition (CELF-4; Semel, Wiig, & Secord, 2003) was used to measure participants’ oral language skills. Core CELF-4 subtests were administered as appropriate according to age to derive receptive and expressive composite scores. Participants’ receptive vocabulary was also evaluated with the Peabody Picture Vocabulary Test–3rd Edition (PPVT-3; Dunn & Dunn, 1997). In addition, participants completed the Letter-Word Identification and Word Attack subscales of the Woodcock-Johnson III Tests of Achievement for Reading (WJ-3; Woodcock, McGrew, & Mather, 2001) to derive Basic Reading (i.e., decoding) scores.

Pretest checklist

During the first experimental session, participants completed a pretest checklist containing 32 rare words to be used in the study (e.g., surfeit, repine), along with 12 common, presumably known words (e.g., hurry, forest). Participants were asked to circle any words that looked at all familiar and to define or write a sentence for words they knew (for a detailed description of the checklist, see Wagovich et al., 2012). The checklist was used to probe for any prior knowledge of the rare words. Performance on common words was included in analyses in an attempt to take into account participants’ general attention and understanding of the task.

Multiple choice posttest

The multiple choice measure contained two parts. The first section assessed the children’s syntactic knowledge of the 32 rare and 12 common words, and the second section focused on general semantic knowledge of the words. All the items relating to syntactic knowledge were alike. Each question asked whether the word was (a) a “person, place, thing, or idea” or (b) an “action word.” A third, “don’t know,” option was provided to discourage random guessing.

The part of the test measuring semantic domain word knowledge included questions that assessed general category information. Fifteen categories, such as “an amount of something,” “saying or hearing something spoken,” “movement,” and so on, were developed to encompass the meanings of all 32 rare words. These categories were used as answer choices. Each category was used as an answer choice 10 to 13 times. The frequency of answer choices used was controlled to the extent possible not to furnish inadvertent meaning cues to participants. A “don’t know” option was provided to discourage guessing.

Three randomized versions of the 88-item multiple choice measure were developed, differing only in the order in which the items appeared. Each participant received each version at least once over the course of the four posttest sessions. Therefore, any test fatigue that children experienced should have affected performance equally across all the rare words.

Stimuli

Rare and common words

The 32 rare words were inserted into written passages (see Wagovich et al., 2015, for a complete list of the rare words). Several “vocabulary builder” sources were consulted to select the rare words used in the present study. Half of the rare words were nouns and half were verbs. All the rare words were two syllables in length and rare enough to not be included as a statistical entry in Carroll, Davies, and Richman’s (1971) word frequency data set. All rare words were consistent with the content of the stories in which they were to be embedded. It was not possible to formally control the richness or informativeness of the contexts in which the words were embedded while still maintaining the naturalness of the texts themselves. However, while inserting the words in the stories, no attempt was made to add contextual cues as to the words’ meanings. In terms of the importance of each word to the passage as a whole, words were, by necessity, less critical to the content of the story, because each could occur in the story only one time. In addition, the 12 common words were chosen from the stories (six nouns, six verbs). Common words had an average Kucera–Francis frequency of 44.8 (SD = 36.6) and an average printed familiarity rating of 531.5 (SD = 53.7) in the Medical Research Council Psycholinguistic Database (Wilson, 1988). Thus, it was judged that participants would be familiar with these words from prior experience.

Stories

Four stories from sixth-grade reading collections served as the passages into which the words were inserted. As described in Wagovich et al. (2012), the stories were modified slightly to be similar in length (from 2,569 to 2,979 words). Readability indices were similar across all four stories, with Flesch–Kincaid grade-level scores (Flesch, 1974) ranging from 4.4 to 4.9. Each of the 32 rare words was inserted only once in one of the four stories, so that each story contained four rare nouns and four rare verbs. Stories were printed on white paper, using 14-point Times New Roman font and 1.5 line spacing. Each story was between 12 and 15 pages in length. For a description of the stories, see Wagovich and Newhoff (2004).

Procedure

All participants attended two initial testing sessions to determine whether they met eligibility criteria for the study. As previously described in Wagovich et al. (2012), during these sessions, held 1 to 3 days apart, the following measures were administered in randomized order: CELF-4, PPVT-3, WJ-3 reading subtests, hearing screening, and nonverbal reasoning screening (SR-DTLA-4). In the first of these two sessions, participants also completed the pretest checklist. Children who qualified for the study attended four additional sessions in which the experiment was administered. During the third, fourth, and fifth sessions (each 2–3 days apart), participants read the same two stories (of four possible, counterbalanced). An examiner sat with each child and asked the child to take his or her time and read silently. The examiner observed the child, but beyond this, did not offer any additional prompts or directives. The reading episodes were intended to resemble, as closely as possible, the natural reading experiences the child might have at school or at home. Toward this end, participants were told in the first session that the study was about children’s reading experiences, so as not to draw attention to the word learning focus. After completion of the study, the experimenter and child discussed the fact that the study was about word learning through reading experiences.

After reading each story, participants were asked to give an oral summary to confirm their overall story comprehension and then were given a short break. Next, the multiple choice posttest was administered to assess syntactic and semantic knowledge of rare and common words. At each posttesting occasion, all participants were tested on all 32 rare words. The 16 rare words that a child encountered in the two stories he or she had read were considered Context words for that child, whereas the 16 words occurring in the two stories the child had not read were considered No Context words for that child. This was done so that word knowledge could be analyzed separately for each participant in both Context and No Context conditions. All participants were also tested on all 12 common words on each occasion. In this way, by the end of the fifth session, participants had read the same two stories three times each and had completed the multiple choice measure three times. The three sessions in which participants completed the multiple choice posttest (i.e., Sessions 3, 4, and 5) will be referred to hereafter as T1, T2, and T3.

Computing Syntactic and Semantic Word Knowledge Scores

Scores on the multiple choice posttest were computed separately for common and rare words. For each word, a correct response to the syntactic item was awarded one point, and a correct response to the semantic item was awarded one point. Performance on common words was included in our models as a covariate, to control for participants’ attention to the task and comprehension of task requirements. For each session, total scores on words participants encountered while reading the stories (Context condition) were computed separately from scores on words seen only during assessment (No Context condition).

Analysis

Analyses focused on the underlying skills associated with word knowledge when context was or was not available. The first step was to explore, on a descriptive level, the language or reading skills associated with overall syntactic and semantic knowledge of the rare words when context was present and when it was absent. To minimize problems with collinearity between participants’ receptive and expressive language scores in subsequent analyses, an overall Total Language standard score was calculated for each participant by averaging the CELF-4 receptive and expressive composite scores for each child. Two-tailed zero-order correlation analyses were used to examine the relations among overall word knowledge scores (measured at T1, T2, and T3 for both Context and No Context conditions) and children’s language and reading scores (CELF-4 Total Language, PPVT-3, and WJ-3 Basic Reading).

Because the study involved repeated measures with the same participants (resulting in multiple scores for each child that were likely to be highly correlated), linear mixed models (SPSS Linear Mixed Model procedure; IBM Corporation, 2013, Version 22.0) were used to explore participants’ growth in word knowledge across additional exposures in the fixed effects models while controlling for individual variation in the random effects models. Separate analyses were conducted to examine participants’ Context and No Context performance. With word knowledge score as the outcome variable, fixed effects were examined for form class (noun, verb), word knowledge type (syntactic, semantic), and time (T1, T2, T3). The primary continuous independent variables of interest were CELF-4 Total Language score and receptive vocabulary as measured by the PPVT-3. Age, WJ-3 Basic Reading scores, and scores on common words were entered as covariates to control for the effects of participants’ age, decoding skill, and attention to task requirements. All main effects were analyzed, as well as two-way (and higher order) interaction effects among all independent variables. Final overall model selection was guided by minimizing the Akaike information criterion (AIC; Akaike, 1974) as well as the Bayesian information criterion (BIC; Schwarz, 1978). Additional details of the statistical analyses are provided in the supplemental materials online.

Results

Descriptive Statistics and Bivariate Analyses

There was a wide range of performance on all standardized language and reading measures: CELF-4 Total Language, WJ-3 Basic Reading, and PPVT-3, as previously reported in Wagovich et al. (2015). Descriptive statistics for these variables are presented in Table 1. Word knowledge scores on the multiple choice posttest also ranged widely across all three experimental sessions, in both Context and No Context conditions. Summary statistics are provided in Table 2.

Table 1.

Descriptive Statistics for Language and Reading Measures.

Test or composite	M	SD	Range
CELF-4^a Total Language	97.5	15.4	72–131
WJ-3^b Basic Reading	98.5	10.8	81–124
PPVT-3^c	107.9	15.8	74–136

Note. Standard Scores M = 100, SD = 15. CELF-4 = Clinical Evaluation of Language Fundamentals–4th Edition; WJ-3 = Woodcock-Johnson III Tests of Achievement; PPVT-3 = Peabody Picture Vocabulary Test–3rd Edition.

CELF-4 (Semel, Wiig, & Secord, 2003); Total Language = mean of receptive and expressive composites. ^bWJ-3 (Woodcock, McGrew, & Mather, 2001). ^cPPVT-3 (Dunn & Dunn, 1997).

Table 2.

Word Knowledge Scores.

Score		Time 1	Time 2	Time 3
Context	M (SD)	38.5 (16.1)	41.3 (17.9)	46.5 (17.5)
Context	Range	6.3–68.8	6.3–75.0	9.4–81.3
No Context	M (SD)	36.4 (13.0)	38.1 (11.7)	40.4 (11.4)
No Context	Range	12.5–62.5	9.4–56.3	15.6–59.4

Note. Scores represent total percentage correct.

Responses on the pretest checklist were examined to assess preexisting knowledge of the rare words. Among the 32 participants, only two children were each able to provide a meaningful sentence for one of the rare words. Thus, it was judged that prior knowledge of these words was negligible. In contrast, participants provided an average of 94.8% (SD = 9.0) correct definitions or sentences for the common words.

As an initial step, all zero-order correlations among word knowledge scores and reading and language measures were explored. Results are provided in Table 3. Scores on all the standardized tests were significantly intercorrelated, and all measures were significantly correlated with total word knowledge scores in both Context and No Context conditions.

Table 3.

Two-Tailed Correlations: Word Knowledge Scores With Language/Reading Scores.

Variable	PPVT-3^d	WJ-3^c Basic Reading	CELF-4^b Total Language	No Context word knowledge score^a
Context word knowledge score^a	.41*	.64***	.56***	.87***
No Context word knowledge score^a	.39*	.53**	.57***
CELF-4^b Total Language	.80***	.68***
WJ-3^c Basic Reading	.60***

Note. PPVT-3 = Peabody Picture Vocabulary Test–3rd Edition; WJ-3 = Woodcock-Johnson III Tests of Achievement; CELF-4 = Clinical Evaluation of Language Fundamentals–4th Edition.

Total score, aggregated across all three posttesting occasions. ^bCELF-4 (Semel et al., 2003); Total Language = mean of receptive and expressive composites. ^cWJ-3 (Woodcock et al., 2001). ^dPPVT-3 (Dunn & Dunn, 1997).

*p < .05. **p < .01. ***p < .001.

Word Knowledge Growth—Overall Findings

To compare factors related to word knowledge growth across conditions, separate Context and No Context linear mixed models were constructed. Initial inspection of residuals revealed no obvious violations of linearity, homoscedasticity, or normality. Overall word learning score was the dependent variable for both models. Form class (noun and verb), word knowledge type (syntactic and semantic), and time (T1, T2, T3) were each entered into the models as fixed factors. CELF-4 Total Language and PPVT-3 scores were initially entered as continuous independent variables. However, in both models, the PPVT-3 failed to account for significant variance, either alone or in interaction with other variables, and was, therefore, excluded from subsequent models for the sake of parsimony. Age, decoding skill (WJ-3 Basic Reading scores), and scores on common words were controlled in all analyses. Examination of the variance–covariance matrix revealed that a first-order autoregressive (AR1) covariance structure was most appropriate for the random effects model. Therefore, an AR1 structure was chosen for tests of random effects. Detailed statistical results of both linear mixed models are provided in the supplementary materials accompanying this article. These results are summarized in the sections that follow.

Context versus no context condition

Results revealed a pattern in which the knowledge accrued in the Context condition was greater than knowledge in the No Context condition. Results of a t test revealed that overall Context word knowledge scores, aggregated across all three times points, were significantly higher than overall No Context word knowledge scores, t = 2.48, p < .05. However, it appears that three exposures were required for participants to gain greater word knowledge of rare words contained in stories, relative to words encountered in isolation. Planned pairwise contrasts at each time point, with Bonferroni correction, revealed that overall Context and No Context word knowledge did not differ at T1 (t = 1.16, p = .26) or at T2 (t = 1.72, p = .10). It was not until T3 that overall Context scores became significantly higher than overall No Context scores (t = 3.13, p < .01).

Growth in word knowledge over time

Results of the separate linear mixed models revealed differing patterns of word knowledge growth over time in Context and No Context conditions. For words encountered in Context, a significant effect of time was found. Follow-up contrasts, with Bonferroni correction, showed that whereas scores at T1 did not differ from scores at T2 (t = 1.59, p = .12), scores at T3 were higher than scores at both T1 (t = 4.05, p < .001) and T2 (t = 3.85, p < .001). Thus, there was growth in word knowledge in the Context condition, but this growth did not become significant until participants had read the stories three times. In contrast, scores in the No Context condition did not differ significantly across any of the experimental sessions, indicating that participants’ word knowledge did not improve with repeated exposure to rare words presented in isolation. These findings are similar to those observed in Wagovich et al. (2015) with the same participants, although the statistical approach used here was somewhat different.

Effects of Language Ability

Context versus no context condition

Significant effects of language ability (CELF-4 Total Language) on overall word knowledge scores were found in both Context and No Context conditions. Above and beyond age, decoding ability, and scores on common words (as a measure of attention and ability to complete the task), participants’ language skills contributed uniquely to their success in gleaning knowledge of the rare words, for words embedded in stories (F = 4.70, p = .04) as well as words encountered only during posttesting (F = 4.47, p = .04).

Growth in word knowledge over time

Although participants with higher language ability gained more overall word knowledge over the course of three exposures, language skill did not appear to affect the pace of word knowledge growth. No significant interaction of time with Total Language was found in either the Context (F = 2.31, p = .10) or the No Context model (F = 0.79, p = .45). Contrary to expectations, higher language skill was not related to a faster rate of word knowledge growth for words presented either with or without surrounding context.

Effects of lexical form class

Language ability showed different patterns of association with performance on nouns and verbs in Context and No Context conditions. For words encountered in Context, language skill contributed similarly to noun and verb scores, as indicated by the lack of significant Total Language × Form class interaction (F = 1.36, p = .25). For words encountered with No Context, however, higher language skill was related to better performance on nouns (t = 3.36, p < .001). Of note, higher language ability was not associated with a specific verb-learning advantage (i.e., in relation to scores on nouns) in either condition.

Syntactic and semantic word knowledge

Language ability was also differentially related to syntactic and semantic word knowledge across Context and No Context conditions. For words presented in Context, a lack of significant interaction between Total Language and word knowledge type indicated that language skill contributed similarly to both types of word knowledge. However, for words presented with No Context, higher language skill was related to greater syntactic (but not semantic) word knowledge (t = 2.63, p = .009).

Discussion

In this study, we explored how language skill might support word knowledge growth during silent reading among school-age children and adolescents. We examined participants’ growth in knowledge of syntactic and semantic aspects of rare words under two conditions: when words were embedded in narrative passages with potential contextual cues to meaning (Context condition) and when words were presented in isolation (No Context condition). Inclusion of a No Context condition allowed us to isolate the effects of encountering rare words within a meaningful written context, above and beyond the effects of repeated exposure to isolated rare words in print.

Overall Findings

Growth in word knowledge

When scores were aggregated across all three posttesting occasions, participants demonstrated greater total syntactic and semantic knowledge of rare words embedded in stories than they did for words encountered only in isolation. However, examination of growth in word knowledge over time revealed that Context scores did not improve until after participants had read the passages three times and did not become significantly higher than No Context scores until after the third reading. Thus, it appears that participants needed at least three exposures to rare words in context for significant learning to occur. This finding is consistent with that of Wagovich and Newhoff (2004), who reported that a single exposure to rare words in passages did not result in significant syntactic or semantic word knowledge growth. Moreover, the present finding that word knowledge did not improve upon repeated exposure when rare words were presented in isolation confirms what is already known about incidental word knowledge growth through both oral exposure and reading: Word learning is a slow and incremental process that requires many encounters across varied contexts for full mapping to take place (Carey, 1978; Swanborn & de Glopper, 1999). In the absence of meaningful context, repeated exposure did not suffice for participants to improve their knowledge of isolated rare words.

Effects of Language Skill

Context and No Context word knowledge

Oral language skill was significantly related to overall rare word knowledge, for isolated words as well as words inserted in passages. The Context finding confirms those of both Steele and Watkins (2010) and Steele (2015), who found that school-age children with typical language performed better than their peers with LLD in gleaning syntactic and semantic knowledge of words during silent reading. It is interesting, although perhaps not surprising, that participants with higher language skills in the current study also outscored their counterparts with lower language skills on rare words that were presented in isolation. In all likelihood, students with greater language ability were able to apply their prior knowledge of phonology, orthography, and morphology to draw more accurate inferences about the syntactic and semantic aspects of words, even when supportive context was not available. Thus, oral language skills may be of vital importance for gaining word knowledge under less-than-ideal conditions as well as in more supportive circumstances.

Receptive vocabulary did not emerge as a significant predictor of word knowledge scores, above and beyond oral language skill. This finding is somewhat at odds with that of Ricketts et al. (2011), who reported that expressive vocabulary was related to word knowledge gained through reading. However, Ricketts et al.’s analysis did not include a broad measure of language ability. Steele and Watkins (2010) found that receptive vocabulary predicted syntactic–semantic word knowledge scores, but only among children with LLD. Standardized vocabulary measures, dependent as they are on crystallized prior knowledge, may be relatively ineffective for predicting the word learning process as it actually occurs (Campbell, Dollaghan, Needleman, & Janosky, 1997; de Villiers, 2004).

Language skill and growth in word knowledge over time

The rate of word knowledge growth across three exposures did not differ according to participants’ language ability. This finding conflicts with that of Steele (2015), who found that children with LLD, but not their typically developing age- or vocabulary-matched peers, benefited from additional exposure. However, our finding is consistent with that reported by Steele and Watkins (2010) indicating that groups of school-age children with LLD and typical language responded similarly to varying levels of exposure to novel words during silent reading. It is not obvious why language skill did not affect the pace of word knowledge growth in the current study. It is possible that skill-based differences would have emerged if participants had been exposed to rare words more than three times. However, participants with lower language skills may have benefited from the repeated presentation of words within the same story contexts. A previous study found that children with lower vocabulary have particular difficulty coordinating information about unfamiliar words across multiple contexts (McKeown, 1985) and, thus, may be more successful in gleaning initial word knowledge when words are encountered repeatedly within a single context. In any case, our finding that word knowledge developed at a similar rate for participants with a wide range of language ability is encouraging. The result suggests that even children with relatively weak oral language skills can achieve meaningful gains in vocabulary through their reading experiences, given appropriate opportunity.

Language skill and lexical form class

For rare words encountered in stories, participants’ language skills were related similarly to performance on both nouns and verbs. When context was available, therefore, students appeared to rely on oral language ability to make inferences about both classes of words. For words encountered without context, however, language skill was related only to performance on nouns. Given that overall scores were higher on verbs than nouns for rare words presented in isolation, participants may not have had to draw as heavily on their language skills to glean information about No Context verbs as they did for No Context nouns. Orthographic or morphological clues particularly associated with verbs may have enabled participants to more readily glean syntactic and semantic information about verbs, even without the benefit of context. Nouns, in contrast, may have posed a greater linguistic challenge when context was not available.

There were no findings to indicate that participants with lower language skills performed more poorly on verbs than on nouns in either condition. One possible explanation for the surprisingly strong verb-learning performance of children with lower language skills is that in the Context condition, verbs may have been more central to overall story comprehension. Participants, especially those with lower language skills, may have been willing to gloss over unfamiliar nouns, but to understand the gist of the stories, they had to gain at least some sense of the verbs’ meanings.

Language skill and syntactic and semantic word knowledge

For rare words encountered in passages, language ability was associated to a similar extent with both syntactic and semantic aspects of word knowledge. Thus, when context was available, participants were able to rely on their oral language skills to infer both grammatical and meaning-related information about the words. However, for rare words encountered only during assessment, language ability was related only to syntactic word knowledge. Without the support of surrounding written context, participants’ understanding of orthography and morphology may have helped them identify words’ form class (i.e., glean syntactic knowledge), but language skill did not confer a similar advantage for interpreting the meanings of isolated words.

Theoretical and Practical Implications

With the continued emphasis in state educational standards on reading as a primary source of content knowledge across the curriculum for students of all ages, it has become more important than ever to understand the factors related to word learning during reading. Analyzing students’ word knowledge separately in Context and No Context conditions enabled us to isolate the variables, including language skill, that contribute to word knowledge growth during meaningful reading experiences. We found that the context provided by narrative passages afforded the necessary level of support for students to make use of underlying oral language skills to glean new word knowledge. In combination with contextual support, language ability was especially important for recognizing the form class of rare nouns and for gathering semantic information about both nouns and verbs. These findings have clear educational implications. Curricula that rely heavily on silent reading as a source of new content-related vocabulary may serve children with higher language skills well. However, such demands are likely to pose stiff challenges for children with lower language ability, who may be less able to make use of written context to infer new word meanings.

Clinical implications, then, are that additional scaffolding of the word learning process will be needed for children with weaker language skills. For example, these children may benefit from strategies that emphasize the use of context clues to derive semantic knowledge of words. Single exposures of words are not generally sufficient to acquire detailed semantic representations of words, but they may be sufficient to develop basic knowledge of a word’s semantic category. With training in deriving word meanings from context, children with weaker language skills may strengthen the ability to derive this basic level of word knowledge through reading. Future research should investigate this possibility by examining the effects of various forms of word learning scaffolding in the development of basic semantic knowledge of a word.

It is important to note that although participants with lower language skill acquired less overall word knowledge than their counterparts with higher language ability, the current study yielded no evidence that word learning during reading is a qualitatively different process for students with low language skill. A sizable number of participants (34%) had receptive or expressive language scores that were at least 1.25 standard deviations below the mean, the generally accepted cutoff for clinical language impairment (Tomblin, Records, & Zhang, 1996). However, there was no indication that participants with low language ability gained word knowledge at a slower rate during reading than their peers with higher skills. Moreover, for words encountered in passages, language ability did not have differential effects on noun and verb learning or on syntactic and semantic aspects of word knowledge growth.

Limitations and Future Directions

Several limitations of the current study should be noted. First, we acknowledge a limiting methodological decision in the design: Because the “context” words were seen in the stories as well as in the assessment phases, children were exposed to these words three more times over the course of the study than the “no context” words, which were only seen in the assessment phases. Thus, it is possible that the difference in exposures could have affected the amount of word knowledge observed between the two conditions. Although we cannot rule out this possibility, it is unlikely that the additional exposures alone were responsible for the additional knowledge that developed over time in the context condition because the assessment targeted syntactic and semantic knowledge, which is less likely to be affected by additional exposures. Nonetheless, a conservative interpretation of our findings is that the provision of context, possibly in combination with the extra exposure provided through the reading experience, contributed to the total learning effect observed in the context condition. A second limitation is that the posttest items for syntactic knowledge, in particular, required some metalinguistic ability. Children had to think about whether words were a “person, place, thing, or idea” or “an action.” This may have affected the children’s ability to demonstrate their syntactic knowledge of the words. However, based on children’s strong performance on the “common word” syntactic and semantic items, we suspect that the metalinguistic load did not substantially impact performance.

Although the sample included students of widely varying language and reading abilities, the sample size was relatively small. Thus, power was sufficient to detect only large effects. Although we acknowledge that with a larger sample our analysis might have detected smaller effects, we argue that from a clinical perspective, large effects are most informative. In addition, the majority of participants were girls, limiting how the findings might generalize to typical classroom settings or to clinical populations with higher proportions of boys. Next, for the posttest, we measured only comprehension of rare words and not expressive word knowledge. We did this for two reasons. First, asking students to generate definitions would have called greater attention to the words, thus detracting from the study’s intended focus on incidental word learning during reading. Second, previous findings suggest receptive measures are more sensitive when assessing early word knowledge growth than expressive measures (e.g., Horton-Ikard & Ellis Weismer, 2007).

Further study is needed to better understand the factors that support incidental word learning during silent reading. The present investigation focused on learner and word-level variables as predictors of word knowledge growth. It is likely that other factors, such as written contextual variables, are also important. Currently, only a few studies have explored how qualities of written context contribute to word learning during silent reading (de Leeuw, Segers, & Verhoeven, 2014; Steele & Watkins, 2010). Our finding indicating that children with relatively weak language skills can acquire new word knowledge through silent reading is promising. Future studies might focus on the ways that specific learner-, word-, and context-level variables interact to affect the likelihood of incidental word learning. Intervention studies focusing on guided learning opportunities to enhance incidental vocabulary growth for children of lower ability (perhaps through the use of dynamic assessment) would provide valuable information about incidental word learning. Future studies might also explore the role of developmentally related cognitive, social, and neurobiological factors that constrain children’s ability to benefit from incidental learning experiences. Given the expectation that school-age children will gain the bulk of their vocabulary incidentally through reading, a better understanding of the factors that maximize such learning for individual children is needed.

Footnotes

Acknowledgements

The study was completed by the first author in partial fulfillment of the requirements for the degree of Master of Health Science, Department of Communication Science and Disorders, University of Missouri. We gratefully acknowledge feedback provided by Judith Goodman in development of the study.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by Grant R03DC006827 awarded to the second author by the National Institute on Deafness and Other Communication Disorders. The first author received funding as a graduate research assistant of the University of Missouri. The second and third authors receive salary as faculty of the University of Missouri. No other financial or nonfinancial relationships exist.

References

Adlof

S. M.

Perfetti

C. A.

(2013). Individual differences in word learning and reading ability. In Stone

Ehren

Silliman

Wallach

(Eds.), Handbook of language and literacy: Development and disorders (2nd ed., pp. 246–264). New York, NY: Guilford Press.

Akaike

(1974). A new look at the statistical model identification. IEEE Transactions on Automatic Control, 19, 716–723.

Anglin

J. M.

Miller

G. A.

Wakefield

P. C.

(1993). Vocabulary development: A morphological analysis. Monographs of the Society for Research in Child Development, 58(10), 1–186.

Biemiller

(2006). Vocabulary development and instruction: A prerequisite for school learning. Handbook of Early Literacy Research, 2, 41–51.

Brown

R. W.

(1957). Linguistic determinism and the part of speech. The Journal of Abnormal and Social Psychology, 55, 1–5.

Cain

Oakhill

(2011). Matthew effects in young readers reading comprehension and reading experience aid vocabulary development. Journal of Learning Disabilities, 44, 431–443.

Campbell

Dollaghan

Needleman

Janosky

(1997). Reducing bias in language assessment: Processing-dependent measures. Journal of Speech, Language, and Hearing Research, 40, 519–525.

Cardoso-Martins

da Silva

J. R.

(2010). Cognitive and language correlates of hyperlexia: Evidence from children with autism spectrum disorders. Reading and Writing, 23, 129–145.

Carey

(1978). The child as word learner. In Halle

Bresnan

Miller

G. A.

(Eds.), Linguistic theory and psychological reality (pp. 264–293). Cambridge: MIT Press.

10.

Carroll

J. B.

Davies

Richman

(1971). The American heritage word frequency book. Boston, MA: Houghton Mifflin.

11.

de Leeuw

Segers

Verhoeven

. (2014). Context, task, and reader effects in children’s incidental word learning from text. International Journal of Disability, Development and Education, 61, 275–287.

12.

de Villiers

J. G

. (2004). Cultural and linguistic fairness in the assessment of semantics. Seminars in Speech and Language, 25(1), 73–90.

13.

Dunn

(1997). Peabody Picture Vocabulary Test-III (PPVT-III) (3rd ed.). Circle Pines, MN: American Guidance Service.

14.

Eyer

J. A.

Leonard

L. B.

Bedore

L. M.

McGregor

K. K.

Anderson

Viescas

(2002). Fast mapping of verbs by children with specific language impairment. Clinical Linguistics & Phonetics, 16, 59–77.

15.

Flesch

R. F.

(1974). The art of readable writing with the Flesch readability formula. New York, NY: Harper & Row.

16.

Gelman

S. A.

Taylor

(1984). How two-year-old children interpret proper and common names for unfamiliar objects. Child Development, 55, 1535–1540.

17.

Gleitman

(1992). A picture is worth a thousand words, but that’s the problem: The role of syntax in vocabulary acquisition. Current Directions in Psychological Science, 1, 31–35.

18.

Hammill

D. D.

(1998). Detroit Tests of Learning Aptitude–4. Austin, TX: Pro-Ed.

19.

Hirsch

E. D.

(2003). Reading comprehension requires knowledge–of words and the world: Scientific insights into the fourth-grade slump and the nation’s stagnant comprehension scores. American Educator, 27(1), 10–29.

20.

Horton-Ikard

Ellis Weismer

(2007). A preliminary examination of vocabulary and word learning in African American toddlers from low socioeconomic status homes. American Journal of Speech-Language Pathology, 16, 381–392.

21.

IBM Corporation. (2013). IBM SPSS Statistics for Windows (Version 22.0). Armonk, NY: Author.

22.

Marinellie

S. A.

Kneile

L. A.

(2012). Acquiring knowledge of derived nominal and derived adjectives in context. Language, Speech, and Hearing Services in Schools, 43, 53–65.

23.

McKeown

M. G.

(1985). The acquisition of word meaning from context by children of high and low ability. Reading Research Quarterly, 20, 482–496.

24.

Nagy

W. E.

Herman

P. A.

(1987). Breadth and depth of vocabulary knowledge: Implications for acquisition and instruction. In McKeown

M. G.

Curtis

M. E.

(Eds.), The nature of vocabulary acquisition (pp. 19–36). Hillsdale, NJ: Lawrence Erlbaum.

25.

Nash

Donaldson

M. L.

(2005). Word learning in children with vocabulary deficits. Journal of Speech, Language, and Hearing Research, 48, 439–458.

26.

Rice

M. L.

Cleave

P. L.

Oetting

J. B.

(2000). The use of syntactic cues in lexical acquisition by children with SLI. Journal of Speech, Language, and Hearing Research, 43, 582–594.

27.

Rice

M. L.

Oetting

J. B.

Marquis

Bode

Pae

S. Y.

(1994). Frequency of input effects on word comprehension of children with specific language impairment. Journal of Speech and Hearing Research, 37, 106–122.

28.

Ricketts

Bishop

D. V. M.

Nation

(2008). Investigating orthographic and semantic aspects of word learning in poor comprehenders. Journal of Research in Reading, 31, 117–135.

29.

Ricketts

Bishop

D. V. M.

Pimperton

Nation

(2011). The role of self-teaching in learning orthographic and semantic aspects of new words. Scientific Studies of Reading, 15, 47–70.

30.

Schwanenflugel

P. J.

Stahl

S. A.

McFalls

E. L.

(1997). Partial word knowledge and vocabulary growth during reading comprehension. Journal of Literacy Research, 29, 531–553.

31.

Schwarz

(1978). Estimating the dimension of a model. The Annals of Statistics, 6, 461–464.

32.

Semel

Wiig

E. H.

Secord

W. A.

(2003). Clinical evaluation of language fundamentals (4th ed.). San Antonio, TX: Psychological Corporation.

33.

Stanovich

K. E.

(1986). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly, 22, 360–407.

34.

Steele

S. C.

(2015). Does language learning disability in school-age children affect semantic word learning when reading? International Journal of Speech-Language Pathology, 17, 172–184.

35.

Steele

S. C.

Watkins

R. V.

(2010). Learning word meanings during reading by children with language learning disability and typically-developing peers. Clinical Linguistics & Phonetics, 24, 520–539.

36.

Swanborn

M. S. L.

De Glopper

(1999). Incidental word learning while reading: A meta-analysis. Review of Educational Research, 69, 261-285.

37.

Swanborn

M. S. L.

De Glopper

(2002). Impact of reading purpose on incidental word learning from context. Language Learning, 52(1), 95-117.

38.

Tomblin

J. B.

Records

N. L.

Zhang

(1996). A system for the diagnosis of specific language impairment in kindergarten children. Journal of Speech, Language, and Hearing Research, 39, 1284–1294.

39.

Wagovich

S. A.

Hill

M. S.

Petroski

G. F.

(2015). Semantic-syntactic partial word knowledge growth through reading. American Journal of Speech-Language Pathology, 24, 60–71.

40.

Wagovich

S. A.

Newhoff

(2004). The single exposure: Partial word knowledge growth through reading. American Journal of Speech-Language Pathology, 13, 316–328.

41.

Wagovich

S. A.

Pak

Miller

M. D.

(2012). Orthographic word knowledge growth in school-age children. American Journal of Speech-Language Pathology, 21, 140–153.

42.

Wilson

M. D.

(1988). The MRC psycholinguistic database: Machine readable dictionary (Version 2). Behavioural Research Methods, Instruments, & Computers, 20(1), 6–11.

43.

Woodcock

R. W.

McGrew

K. S.

Mather

(2001). Woodcock-Johnson III Tests of Achievement. Itasca, IL: Riverside Publishing.