Tracking the acquisition of orthographic skills in developing readers: Masked priming effects

https://doi.org/10.1016/j.jecp.2007.01.006Get rights and content

Abstract

A masked priming procedure was used to explore developmental changes in the tuning of lexical word recognition processes. Lexical tuning was assessed by examining the degree of masked form priming and used two different types of prime–target lexical similarity: one letter different (e.g., rlay  PLAY) and transposed letters (e.g., lpay  PLAY). The performance of skilled adult readers was compared with that of developing readers in Grade 3. The same children were then tested again two years later, when they were in Grade 5. The skilled adult readers showed no form priming, indicating that their recognition mechanisms for these items had become finely tuned. In contrast, the Grade 3 readers showed substantial form priming effects for both measures of lexical similarity. When retested in Grade 5, the developing readers no longer showed significant one letter different priming, but transposed letter priming remained. In general, these results provide evidence for a transition from more broadly tuned to more finely tuned lexical recognition mechanisms and are interpreted in the context of models of word recognition.

Introduction

Children’s capacity to recognize written words develops with remarkable speed. By the end of the second year of high school, adolescents reading English will typically recognize more than 80,000 words (Adams, 1990). How does this phenomenal capability develop? It is widely accepted that, at a basic level, beginning readers must learn a set of correspondences between the letters or graphemes of written words and the phonemes of spoken words, broadly referred to as alphabetic skills. However, to become accurate and efficient readers, at least of a deep orthography such as English, children ultimately need to acquire a rapid and flexible word recognition system that allows immediate identification of both regular and irregular words. It is this orthographic stage of word recognition that is typically seen as being the hallmark of skilled reading (e.g., Frith, 1985, Perfetti, 1992, Share, 1995).

Although an orthographic stage of word recognition is widely posited in developmental theories, relatively little is known about how children reach this stage or how the process of its acquisition is to be characterized. In recent work, we have attempted to explore this question by using the masked priming technique to tap the acquisition of rapid and automatic word recognition processes in developing readers (Castles et al., 2003, Castles et al., 1999, Davis et al., 1998). This procedure involves briefly presenting participants with a letter string (the prime) and then presenting them with a second letter string (the target) and asking them to perform some task on it (e.g., naming, lexical decision). Facilitation is found to occur when the prime is orthographically the same as the target (e.g., cat  CAT), referred to as repetition priming, or in some cases when it is very similar to it (e.g., cal  CAT), referred to as form priming.

It is generally argued that the prime improves performance in these circumstances by activating the lexical representation for the target word and therefore assisting its processing in the subsequent task (Forster, Mohan, & Hector, 2003). There are several reasons for believing this to be the case. First, participants are generally unaware of having seen the prime, so any effects of explicit conscious processing are minimized (Forster & Davis, 1984). Indeed, if the duration of a word prime is increased so that participants can report it, no form priming is obtained (Humphreys, Evett, Quinlan, & Besner, 1987). Second, because targets are displayed for at least 500 ms, the effects are unlikely to be due to visual integration (Davis & Forster, 1994). Finally, priming is typically small or nonexistent when the target is a nonword (Forster, Davis, Schoknecht, & Carter, 1987; but see Bodner and Masson, 1997, Bodner and Masson, 2001, Masson and Bodner, 2003). If the effects were occurring at other than an orthographic level, equal facilitation would be expected for nonwords as for words.

Masked repetition and form priming effects of various kinds have been widely observed in skilled adult readers (for a review, see Kinoshita & Lupker, 2003). More important for the current purposes, such effects have now been found in children in the process of learning to read. For example, 9-year-olds reliably show strong repetition priming effects for high-frequency words that are present in their written vocabularies (Davis et al., 1998, Pratarelli et al., 1994), and Castles and colleagues (1999) reported similar-sized effects in children as young as 7 years. With this basic finding established, it becomes possible to use the masked priming procedure to explore aspects of word recognition in developing readers in more depth by manipulating aspects of the prime and comparing the pattern of priming shown by children at various stages of reading acquisition with that shown by skilled adult readers.

Castles and colleagues (1999) took this approach in exploring the effect of neighborhood size (N) on form priming in developing readers. Neighborhood size is a broad metric of the similarity of a word to other words and is typically operationalized as the number of other words that can be created from a particular word by changing just one letter in any position. The word clap, for example, is a high-N word because 10 other words can be created from it by changing one letter (e.g., slap, flap, clip, clam), whereas bird is a low-N word because only 1 other word can be created from it by changing only one letter (i.e., bind). Adults have consistently been shown to produce a differential pattern of masked form priming for high- and low-N words: A one letter different form prime will produce facilitation in responding to a low-N word (e.g., lird  BIRD), but such a prime will not produce priming for a high-N word (e.g., blap  CLAP) (Forster et al., 1987). In contrast, Castles and colleagues (1999) found that developing readers in Grades 2, 4, and 6 showed form priming from one letter different primes for all words, regardless of whether they were high or low N.

Castles and colleagues (1999) interpreted this finding in terms of a modification in the tuning of the automatic word recognition system in response to growth in the size of the written vocabulary. Early in word recognition development, they argued, the system can afford to be fairly broadly tuned and to accept similar, but not identical, inputs as candidates for a target word. This is because, at this early stage, many of the similar-looking competitors of the word are not yet in the reader’s vocabulary, and so gains in efficiency can be achieved by adopting a loose criterion without compromising accuracy. As written vocabulary grows, however, the system must adapt to the presence of many more similar competing words in the lexicon in the case of high-N words and so must tighten up the input criterion so as to maintain maximum accuracy. As such, they proposed that the developing word recognition system responds to ongoing changes in the recognition requirements of particular words at an item-based level (for a similar conceptualization of word recognition development as occurring in an item-based fashion, as opposed to a stage-based fashion, see Share, 1995). We refer to this theory henceforth as the lexical tuning hypothesis.

In the current study, we sought to extend the findings of Castles and colleagues (1999) in two ways with the aim of providing further support for the lexical tuning hypothesis. First, we attempted to address an anomaly in their findings regarding the magnitude of form priming for high-N words at different grade levels. Besides predicting a difference between developing readers as a group and adults in the magnitude of priming for high-N words, Castles and colleagues expected to observe a gradual decrease in priming from Grade 2 to Grade 6, reflecting the adaptation of the system to the children’s growth in written vocabulary over this time. This was not found; significant form priming for high-N words was observed even in the Grade 6 readers. However, one limitation of this study was that it was a cross-sectional design; different children were tested at each grade level, and there was considerable variation in reading ability, as well as in the amount of priming, across the different age groups. Therefore, it is possible that this variability masked the presence of any modulation in form priming effects as a function of growth in vocabulary size; indeed, when we examined a subset of the Grade 6 readers who were better readers and who showed less variability in their performance, the magnitude of priming for high-N words was found to be much smaller. In the current study, we attempted to address this problem by first comparing adults with developing readers and then examining the change in priming for those same developing readers at a later date. The same children were assessed on their one-letter-different form priming for high-N words at two points in time: in Grade 3 and again in Grade 5. We hoped that this design would reduce between-group variance and maximize the opportunity to observe subtle changes in the tuning of children’s word recognition mechanisms over time.

The second aim of this study was to explore another index of the precision of the tuning of children’s word recognition mechanisms so as to build on and elaborate the findings for masked one letter different form priming of high-N words. Here we focused not on the coding of letter identity, as in the one letter different priming effects, but rather on the coding of letter position within words. As children learn more and more words, the position of any particular letter within a word becomes more critical to its successful identification; for example, the position of the letters in the word tap needs to be coded more precisely once the words pat and apt must also be identified. Therefore, once again, based on the lexical tuning hypothesis, we expected to see a pattern of development from a fairly loosely tuned system for the coding of the position of letters within words toward a more precisely tuned system.

No research to date has explored masked priming with transposed letter primes (e.g., hopsital) in children. However, such studies have been conducted in adults. Here the results have been somewhat mixed. Several studies report significant priming from nonword transposed letter primes (e.g., anwser) to lexical decisions on target words (e.g., ANSWER) (Forster et al., 1987, Perea and Lupker, 2003, Schoonbaert and Grainger, 2004). However, Andrews (1996) reported no such facilitation on naming times for similar items. It may be that, for adults, the magnitude of priming from transposed letter primes is modulated by factors such as word length and lexical density, which varied considerably across these three studies. In the current study, we selected short, high-frequency, and high-N words with the expectation that adult tuning for letter position for these items would likely be quite precise and, therefore, that priming from transposed letter primes would be minimal. At the very least, we expected that the magnitude of priming would be smaller for these items in the adults than in our developing reader sample.

In summary, in the current study, we explored one-letter-different and transposed letter form priming in a group of adults and in a group of children when they were in Grade 3 and then, 2 years later, in the same children when they were in Grade 5. For both types of prime, we predicted that the Grade 3 readers would show a greater magnitude of priming than would the adults. At follow-up, we expected that the magnitude of priming for the participants in Grade 5 would be less than that shown in Grade 3 and that the magnitude of priming might fall between that shown by the Grade 3 readers and that shown by the adults. Such a pattern, if found, would provide support for the lexical tuning hypothesis.

Section snippets

Participants

The adult participants were 24 undergraduate and graduate students from the University of Melbourne (16 women and 8 men, mean age = 18 years 10 months, SD = 13 months). All spoke English as their first language.

The child participants were 23 Grade 3 students from a Melbourne primary school1 (14 girls and 9 boys, mean age = 8 years 6 months, SD = 5 months). They comprised one

Results

Table 1 presents summary statistics for the adults and the children at each grade level for the substitution, transposition, and control priming conditions. These data were analyzed in two separate sets of analyses. First, the pattern of priming for the adults versus the Grade 3 readers was compared in a set of between-group ANOVAS. Following this, the results for the follow-up testing of the participants in Grade 5 were analyzed separately. There were two reasons for dealing with the Grade 5

Discussion

We begin this section by summarizing the results first for the adults and Grade 3 readers and second for the follow-up testing of the participants in Grade 5. We then consider how the findings might be interpreted in terms of theoretical models of the developing word recognition system.

References (34)

  • C. Davis et al.

    Masked homophone and pseuodohomophone priming in children and adults

    Language and Cognitive Processes

    (1998)
  • C. Davis et al.

    Masked orthographic priming: The effect of prime–target legibility

    Quarterly Journal of Experimental Psychology A

    (1994)
  • L.C. Ehri

    The development of the ability to read words

  • K.I. Forster

    Accessing the mental lexicon

  • K.I. Forster

    The pros and cons of masked priming

    Journal of Psycholinguistic Research

    (1998)
  • K.I. Forster et al.

    Repetition priming and frequency attenuation in lexical access

    Journal of Experimental Psychology: Learning, Memory and Cognition

    (1984)
  • K.I. Forster et al.

    Masked priming with graphemically related forms: Repetition or partial activation?

    Quarterly Journal of Experimental Psychology

    (1987)
  • Cited by (132)

    View all citing articles on Scopus
    View full text