The relationship between knowing a word and reading it aloud in children’s word reading development

Abstract

This experiment examined the item-level relationship between 7-year-olds’ ability to read words aloud and their knowledge of the same words in the oral domain. Two types of knowledge were contrasted: familiarity with the phonological form of the word (lexical phonology), measured by auditory lexical decision, and semantic knowledge, measured by a definitions task. Overall, there was a robust relationship between word knowledge and reading aloud success. The association was stronger when words contained irregular spelling–sound correspondences. There was no evidence that a deeper or more semantic knowledge of words was more closely related to reading aloud success beyond the association between reading success and familiarity with the phonological form of the same words. This finding is not compatible with models that see semantics as contributing directly to the reading aloud process, at least during the relatively early stages of reading development. More critical was whether or not a word was considered a lexical item, as indexed by auditory lexical decision performance.

Introduction

The question of how children learn to read words has generated considerable research interest, resulting in a rich understanding of some aspects of word reading development. Alphabetic decoding—the ability to translate a printed word into its spoken form—clearly plays a pivotal role in early reading development, underpinned by developments in children’s phonological skills (e.g., Ehri, 2005, Share, 1995, Share, 2008b). Equally clear, however, is that factors other than alphabetic decoding and phonological skill must also contribute to word reading development, especially as children get older and become more skilled at reading (e.g., Castles and Nation, 2006, Scarborough, 2005). One factor that might be important is knowledge about the meanings of words. The lexical quality hypothesis (Perfetti, 2007, Perfetti and Hart, 2002) states that effective reading demands high-quality lexical representations that integrate information about form (phonology, orthography, and grammatical class) and meaning (semantics). This hypothesis remains open to the mechanisms that might underpin how semantic knowledge becomes integrated with form during the development of high-quality representations. Some years ago, Plaut, McClelland, Seidenberg, and Patterson (1996) suggested that semantic factors provide a direct contribution to children’s word reading development. Surprisingly, some of the key predictions of their model have not been tested against data from children learning to read. The overall aim of the current investigation was to assess the hypothesis that children’s knowledge of word meanings in the oral domain contributes to their ability to read words aloud. We begin by describing Plaut and colleagues’ model and reviewing some evidence from children that is broadly consistent with its framework before introducing the specific hypotheses evaluated here.

Plaut and colleagues’ (1996) model, the triangle model, is a connectionist one in which lexical information is represented in sets of distributed subsymbolic codes representing lexical attributes (semantic, phonological, and orthographic). Reading aloud follows from the activation of codes from visual input and is accomplished via two pathways working in parallel: a phonological pathway comprising connections between orthography and phonology and a semantic pathway comprising mappings among semantic, phonological, and orthographic representations. Reading aloud is considered to involve both pathways and all types of representation regardless of the lexicality, familiarity, or frequency of the items being read. However, because the phonological pathway is more direct, it is thought to be faster and, therefore, to contribute more to reading aloud than the semantic pathway. Importantly, although the semantic pathway is always active, contributing some information to the reading process, its contribution becomes more important when the phonological pathway is compromised, for example, when reading words that have inconsistent mappings between orthography and phonology. Consistent with this, semantic involvement in reading aloud has been demonstrated most clearly when adults are asked to read inconsistent words (e.g., McKay, Davis, Savage, & Castles, 2008).

Is there any evidence that semantic factors may contribute to children’s word reading skill? One perspective on this question is to note that when children come to the task of learning to read, many of the words they encounter in print are familiar in their oral language. Thus, it is reasonable to ask whether this knowledge influences the ease with which children learn to read. Consistent with word knowledge playing a role in reading development, a number of investigators have shown that individual differences in oral vocabulary knowledge are associated with reading skill. For example, Nation and Snowling (2004) found that word knowledge accounted for unique variance in children’s word reading measured concurrently at 8 years of age and longitudinally when the children’s reading was reassessed 5 years later at 13 years of age, even after variance associated with decoding (nonword reading) and phonological skills was taken into account. If we assume that oral knowledge provides a reasonable proxy for the skills and information that contribute to the semantic pathway, Nation and Snowling’s longitudinal study provides evidence consistent with semantic factors contributing to the development of word reading, as suggested by Plaut and colleagues (1996). However, the nature of this contribution is far from clear. Plausibly, a relationship between vocabulary knowledge and reading development may reflect a general consequence of children who are good or poor at one thing tending to be good or poor at other things. Yet the triangle model assumes a much more specific relationship, with word knowledge playing a more active role at an individual word-by-word level. The first aim of our experiment was to test this assumption. If it is correct, there should be an item-level relationship between word knowledge in the oral domain and reading aloud; that is, an individual’s knowledge of the meaning of words should predict his or her ability to read those words.

A second prediction stemming from the triangle model is that the relationship between word knowledge in the oral domain and reading aloud ought to be strongest when children are asked to read words that have inconsistent mappings between orthography and phonology. Although the semantic pathway contributes to the pronunciation of all words regardless of lexicality or regularity, it is argued to be most important (and most manifest) when reading irregular words because these words are read less successfully by the phonological pathway. Once again, there is evidence from children’s reading that is broadly consistent with this prediction. Goff, Pratt, and Ong (2005) found that receptive vocabulary showed a stronger correlation with irregular word reading (r = .53) than with nonword reading (r = .28) in 10-year-olds. Bowey and Rutherford (2007) observed exactly the same pattern in a group of 13-year-olds, with receptive vocabulary correlating .57 with irregular word reading but only .39 with nonword reading. A more extreme pattern has emerged when oral knowledge is assessed via an assessment of expressive vocabulary. Both Ouellette (2006) and Ricketts, Nation, and Bishop (2007) found that expressive vocabulary predicted significant variance in irregular word reading but not in nonword reading. Taken together, these findings suggest that any item-level relationship between knowledge in the oral domain and reading aloud should be stronger for items that contain irregular spelling–sound relations.

Our third aim was to address the following question: If oral vocabulary knowledge does relate to word reading ability at an item-level, “how much” knowledge is needed? Do children need to have a deep appreciation of the meaning of the word, or does a sense of familiarity suffice? The triangle model is not helpful on this point because “semantics” has not been implemented in a psychologically plausible way. It is, however, a very important question because it gets to the heart of what exactly is meant by a semantic contribution to word reading.

Ouellette (2006) drew a useful distinction between the breadth of children’s oral vocabulary and their depth of vocabulary knowledge. This distinction stems from theoretical work in psycholinguistics that describes the lexicon as a store of phonological word forms that are separate from, but heavily connected to, semantic representations (e.g., Levelt, Roelofs, & Meyer, 1999). Ouellette (2006) found that the two constructs of breadth and depth showed a differential relationship with different aspects of reading. Not surprisingly, vocabulary depth was related to reading comprehension (see Braze, Tabor, Shankweiler, & Mencl, 2007, and Tannenbaum, Torgesen, & Wagner, 2006, for similar findings). More relevant to the current investigation was the finding that vocabulary breadth, but not depth, was related to nonword reading, leading Ouellette to conclude that the relationship between decoding and oral vocabulary is primarily a function of the size of the receptive (phonological) lexicon. A different pattern was seen for irregular word reading, sharing a unique relationship with both vocabulary breadth and depth. Thus, in addition to the size of the phonological lexicon, Ouellette’s (2006) findings suggest that the depth of semantic knowledge in the oral domain places an additional constraint on, or offers additional support to, irregular word reading. This sits comfortably with the idea that the semantic pathway draws on a relatively “deep” level of meaning beyond a mere sense of familiarity.

The distinction between vocabulary breadth and vocabulary depth is logical and theoretically motivated. Critical, however, is how the two constructs are operationalized. Ouellette (2006) used four subtests from the Test of Word Knowledge, a standardized battery (Wiig & Secord, 1992). Two subtests were deemed to measure vocabulary breadth: expressive vocabulary, where children provided the names for depicted objects, and receptive vocabulary, where children selected the appropriate pictures to match spoken words. The two other subtests were considered as assessments of vocabulary depth: word definitions, where children provided an oral definition for a set of words, and synonyms, where children selected a synonym from an array of pictures that match each spoken word. Unfortunately, the distinction between vocabulary breadth and depth is clouded by a number of factors being confounded, including the complexity of task demands, receptive versus expressive skills, and (most crucially) semantic knowledge. Single-word vocabulary tests certainly reflect vocabulary breadth, but to name pictures and to match words to pictures also requires some degree of semantic knowledge. Similarly, definitions and synonyms provide assessments of depth of knowledge, but these are not necessarily independent of vocabulary breadth. This substantial overlap limits the conclusions that can be drawn from Ouellette’s (2006) findings concerning the role of vocabulary knowledge in word reading development.

To directly address the question of how much knowledge is needed to support word reading, we made a distinction between two aspects of word knowledge: lexical phonology, defined as children’s familiarity with the phonological form of a word, and semantics, defined as children’s understanding of the meaning of a word. Lexical phonology was measured using auditory lexical decision, where children heard an item and needed to decide whether it was a word or a nonword. In psycholinguistic studies with adults, auditory lexical decision is generally considered to tap lexical processes rather than semantic processes (Goldinger, 1996). Following this, we asked children to provide definitions of the same words, allowing us to index semantics or depth of vocabulary knowledge. If depth of knowledge is important, we would expect that definition scores would serve as a better predictor of word reading, especially irregular word reading, than lexical phonology alone.

In summary, previous work has established a relationship between oral vocabulary knowledge and learning to read words. We aimed to elucidate the nature of this relationship by asking whether it holds at an item level, as predicted by the triangle model. If so, we would expect a close relationship between children’s knowledge of words in the oral domain and their ability to read aloud the same words successfully. The triangle model also predicts that the role of semantics should be more manifest when the words to be read are irregular or inconsistent. We tested this prediction by examining whether the relationship between word knowledge and word reading is moderated by orthographic regularity. Finally, by comparing performance for two tasks that index differing amounts of knowledge (lexical phonology and semantics), we also assessed the issue of how much knowledge is needed to support word reading.