Abstract
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.
Children with language impairments (LI) have difficulties learning new words in oral discourse that have been attributed to their problems with phonology, semantics, and syntax. Typical language learners are able to acquire some aspects of word meaning from just a few exposures to new words in the context of running dialogue (Rice & Woodsmall, 1988). In contrast, children with LI are not always as proficient as their typical peers in learning words incidentally (Rice, Oetting, Marquis, Bode, & Pae, 1994), and, consequently, have difficulty producing new words after limited exposures and learning the semantic features of new words (Alt, Plante, & Creusere, 2004). Furthermore, children with LI are less sensitive to syntactic cues that provide information about new word meanings (Rice, Cleave, & Oetting, 2000). Words from certain grammatic classes, especially verbs, have been shown to pose significant difficulty (Windfuhr, Faragher, & Conti-Ramsden, 2002).
Less is known about how children acquire new words when reading text. Studies of typically developing children have shown that that when they first encounter a word in text (Nagy, Herman, & Anderson, 1985), they may only learn to recognize the new word as a real word rather than a nonsense word. In fact, after just one reading, children may actually map some incorrect information to a word. Subsequent encounters with the words should help to refine the mental representations of words by adding correct details and eliminating incorrect details. Several text factors influence word learning in a positive way for typically developing children:
Words that occur more frequently are more easily learned (Jenkins, Stein, & Wysocki, 1984).
Concrete, highly imaginable nouns are easier to learn (Stahl, 2003).
When the immediate context provides clues to the words’ meanings, children are more likely to pick up additional word knowledge (Carnine, Kameenui, & Coyle, 1984).
Children with LI have a higher probability of developing a reading disability (e.g., Carroll & Snowling, 2004; Catts, Fey, Tomblin, & Zhang, 2002). In a study of second and fourth graders, children with LI scored significantly lower than children with typical language on tasks of word recognition and reading comprehension, the two main components of reading ability (Catts et al., 2002). In second grade, nearly 53% of the children with LI had a reading impairment, and in fourth grade more than 48% of the children with LI had a reading impairment. Looking at the relationship a different way, about 19% of the children with dyslexia also had LI in second, fourth, and eighth grades. Although differing in their conceptualization of the underlying causes of dyslexia and LI, several models of reading and language ability (e.g., Bishop & Snowling, 2004; Catts & Kamhi, 1999) consider that semantics, syntax, morphology, and discourse-level skills (nonphonological skills or listening comprehension) contribute to reading separately from print skills (phonological skills or word recognition). These models characterize poor readers based on their relative strengths and weaknesses in the two domains—word recognition (decoding abilities) and comprehension. Children with LI, who as a group experience difficulty with semantics, syntax, morphology, and discourse processing, can be expected to have problems with reading comprehension with or without concomitant problems in word recognition. Problems with comprehension have been shown to create difficulty for children with LI when they need to learn word meanings from the context of the text. Studies of word learning during reading have been conducted with children who are typically developing or children with primary reading problems. Considering the word learning problems that children with LI experience in the spoken domain, the high probability of children with LI having concomitant problems with reading and the importance of learning word meanings during reading, research into word learning during reading for children with LI is needed. Much of the current research has focused on word learning in the spoken domain; yet for older school-aged children, word learning is not confined to spoken language. Children also gain understanding of new words that are presented in written form. This investigates word learning for older children during reading.
School-aged children with language-learning disability (LLD) appear to have a multidimensional source of word learning weakness. They often have poor spoken vocabulary and impaired word learning ability. In addition, they have weaker phonological awareness (Catts, 1993) and morphological skills (Windsor, Scott, & Street, 2000), which may negatively impact their reading success. With lower reading success, children with LLD may be exposed to fewer words during reading, which would negatively affect their vocabulary through mere lack of exposure. Many of the same weaknesses in spoken lexical learning would be expected to occur during reading. Even if children with LLD were exposed to the same amount of written text, they would likely learn fewer words due to deficits in underlying word learning ability. If speech-language pathologists (SLPs) and educators are to provide appropriate intervention programs, there is a need for better understanding of the nature of word learning deficits for children with LI and the conditions under which their word learning improves. The primary objective of this study was to examine the ability of children with LLD to learn word meanings during reading. The following primary questions were posed:
Do children with LLD and same-aged peers differ in their ability to define and identify basic word meanings encountered during reading?
Do text variables (position of informative context, number of exposures to target words, and part of speech) affect word learning of children with LLD and same-aged peers?
Do contextual clues affect children’s abilities to provide correct definitions for novel words?
Study
Participants
Participants include 15 children with LLD (9;11–11;5 years) and 15 typically developing age-matched children (9;9–11;9 years). All participants were assessed for language, reading, and nonverbal cognitive ability. Receptive vocabulary ability was assessed using the Peabody Picture Vocabulary Test–4 (Dunn & Dunn, 2007), and grammatic ability with the Concepts and Directions and Formulated Sentences subtests of the Clinical Evaluation of Language Fundamentals, Fourth Edition (CELF-4; Semel, Wiig, & Secord, 2003). Nonverbal cognitive ability was screened using the Test of Nonverbal Intelligence, Third Edition (TONI-3; Brown, Sherbenou, & Johnsen, 1997). To measure fluency and comprehension of reading, the Gray Oral Reading Tests, Fourth Edition (GORT-4; Wiederholt & Bryant, 2001) were given. The two groups differed on all language and reading standardized scores. Although children from both groups scored within the average range on the TONI-3, a significant group difference was found. To control for group differences in nonverbal intelligence, TONI-3 scores were entered as a covariate in all subsequent analyses.
Reading Passages
Original reading passages were written for the present study. The readability of the passages was measured by the Flesch–Kincaid readability formula, calculated automatically in Microsoft Word by measuring the average number of words per sentence and average number of syllables per word. The readability of three passages was at a first-grade level; readability of the fourth passage was at second-grade level. The passages were approximately one page in length, double spaced, without illustrations. The mean number of words per passage was 381, with a range of 281 to 423.
The passages reflected themes that were expected to be familiar to most children: My Family, My Pets, My School, and My Birthday. The passages were written such that the context provided information about each target word; however, there were no devices used to draw attention to the informative context (e.g., “For example” or “This word means . . .”). Instead, the participants were required to infer, or draw conclusions about, the meanings of the words from the passages. Twenty nonsense target words were generated. There were five target words per reading passage, and a total of four reading passages. Ten of the target words were nouns and 10 were verbs. Nonsense words included nouns that were singular and verbs that had bare stems, to keep morphology consistent and to avoid morphological cues for certain words. The target words were one syllable in length, in CVC, CVVC, CCVC, CVCC, or CVCe word forms. The nonsense words followed the orthographic rules of the English language, and were easily decodable for 7- to 8-year-old readers. The meanings of the target words were synonyms for concepts that were likely to be known to the participants. Thus, children were not learning new words for new concepts, but were learning new words for known concepts, a task that older children often face as their vocabulary grows in depth and breadth.
In the reading passages, words occurred in two Rate conditions (low rate, high rate) and two Position conditions (adjacent context, nonadjacent context). Thus, there were five words in each of the following conditions: low rate-adjacent context, low rate-nonadjacent context, high rate-adjacent context, and high rate-nonadjacent context. Each reading passage contained a combination of word conditions. Target words in the low rate condition were presented twice, and in the high rate condition were presented 5 times. In the adjacent context, explicit information describing the target word was positioned in the beginning of the sentence immediately following the target word. In the nonadjacent condition, the explicit informative context was separated from the target word by one neutral sentence. Each target word was placed as the last word in the sentence in which it first appeared. Explicit informative context appeared only once, after the first presentation of each target word. The following is a first paragraph from one of the passages: My school is too far away to walk to. I get there on a
Assessments
Oral definitions
Students were shown a target word written on a 4 6 note card and asked, “Can you say this word?” Responses were phonetically transcribed. Next they were asked, “Can you tell me what X means?” Additional prompts were provided as needed. If students gave an incomplete response, they were told, “That’s part of it. Can you tell me something else about X?” If a vague definition was given, they were asked, “Can you tell me more or give me an example?” If the student still refused to answer or did not know, they were asked, “Does X remind you of anything?” Definitions were scored on a 4-point scale, for example, for the word
No correct information, for example, “I don’t know”
Vague response, for example, “what he does”
Incomplete response, for example, “play”
Complete response, for example, “a running game he plays with his friends”
Multiple choice
Each target word was presented in a phrase, “What is the closest meaning for the word X?” Four choices followed, including (a) correct syntactic category and correct semantic information, (b) correct syntactic category and incorrect semantic information, (c) incorrect syntactic category and incorrect semantic information, (d) none of the above. For example, the meaning for the word zear was a game similar to bowling. The options included (a) to bowl, (b) clean up, (c) fizzy, (d) none of the above. The responses were scored as follows: zero points for incorrect syntactic category and incorrect semantic information and “none of the above” choices, one point for correct syntactic category and incorrect semantic information, two points for correct syntactic category and correct semantic information.
Procedure
All students read all four stories, but in different orders. Assistance was given for decoding individual words if students asked for help. No words were defined. Next, they were asked to provide oral definitions of the target words and then to complete the multiple choice task.
Results
Overall word learning
Children with LLD performed more poorly than their same-aged peers on both oral definition and multiple choice tasks. On the oral definition task, the mean scores and standard deviations for the LLD group showed that most children were able to express an oral definition for very few of the target words. Children with LLD showed complete learning of less than one fifth of the words. The oral definition task was also difficult for the typical children, but they outperformed the children with LLD and showed complete learning of about half of the words. The authors propose several explanations for these results:
The texts were easily decodable by students with LLD; however, as a group, the children with LLD did show lower reading fluency. Even if the children with LLD were able to decode the passages, they may have allocated more cognitive resources to the decoding process (Just & Carpenter, 1992), leaving fewer cognitive resources for comprehending the passage and interpreting word meaning.
Current vocabulary knowledge may have influenced children’s performance on the word learning tasks. Children with stronger vocabulary knowledge may have been able to understand the other words in the passage and use that knowledge to infer the meanings of the target words.
Children with LLD have also shown problems producing correct definitions, which may have deflated their oral definition performance. Children with LLD lag significantly behind their peers in their ability to provide mature definitions (Marinellie & Johnson, 2002). So, they may have known more than they were able to produce in an expressive definition task. For this reason, a multiple choice task was included, which may have given a more accurate comparison of word learning ability. Children with LLD also scored lower than their typical peers on the multiple choice task. Thus, problems producing definitions cannot completely explain their poorer performance.
Context effects
Number of presentations: Students tended to learn more words in the high rate condition, but this difference was not significant. The higher number of presentations did not help students with LLD close the word learning gap with typically developing students. It may be that students with LLD required more than five exposures to close the gap.
Position of informative context: For typical students, word learning was similar for both adjacent and nonadjacent context. Students with LLD exhibit a lower number of correct responses on the nonadjacent context item, but his difference was not significant. The nonadjacent informational sentences were separated from the target word by only one sentence. One-sentence separation may not have been far enough away to see a position effect.
Part of speech: In some word learning studies, students exhibited more difficulty learning verbs than nouns. In this study, both groups of students performed similarly on learning nouns and verbs. This may have been because the words had no syntactic markings, that is, no noun had singular or possessive markers and no verb had a tense or agreement marker.
Contextual cues and oral definition performance
The contextual clues were intended to help children recall word meanings, but not provide enough information for children to produce a completely correct response. More typically developing children received better oral definition scores given contextual clues than students with LLD. This result raises the possibility that the typically developing children had inferred the correct word meaning during reading, but required a contextual clue to recall the meaning during the oral definition task. If children with LLD also had more difficulty with recalling, rather than inferring, they should have shown similar improvement given contextual clues. However, the contextual clues did not help children with LLD to the same degree, which may be indicative of difficulty with inference of word meaning.
Implications
Several relevant clinical implications can be drawn from the findings of the study:
Children with LLD do not show the same propensity for learning the meanings of words from reading as typically developing peers. Thus, they require more direct instruction in vocabulary learning strategies.
Successful vocabulary interventions for typical children include multiple exposures to words across multiple contexts to encourage deep processing of word meanings. Students with LLD require even more exposures. Clinicians may accommodate these needs by using classroom vocabulary items, preteaching core curricular vocabulary, and collaborating with classroom teachers to provide contextually-rich, multiple exposures to target vocabulary items.
Children with LLD showed difficulty inferring word meanings from context, as shown by their poorer performance on the multiple choice task and more limited benefit from contextual cues. Children with poor reading comprehension ability and low vocabulary skills can be taught to use context to generate word meanings (cf. Schatz & Baldwin, 1986) by introducing the text context, reading and rereading the passage, focusing on contextual word clues, and developing, revising, and summarizing their hypotheses of word meanings.
Students obtained higher scores when a dynamic assessment was employed. For example, after reading a passage containing novel words, a student’s proficiency could be determined through a hierarchy of tasks including (a) oral definition without assistance, (b) oral definition with questions to elicit more information, (c) oral definition with a contextual clue, and (d) choosing the correct definition from a field of possibilities.