Development of reading processes: Orthographic and morphological influences

Development of reading processes: Orthographic and morphological influences

First Author: Peter F. de Jong -- University of Amsterdam
Keywords: Word recognition processes, Orthographic Knowledge, Morphological processing, Rapid naming
Abstract / Summary: 

During reading development the processes underlying word identification tend to change as orthographic and morphological knowledge of words is acquired. In this symposium researchers from five different labs consider how reading processes change during reading acquisition and present innovative methods to detect such changes. In some studies tightly controlled experimental paradigms are used to determine how orthographic or morphological knowledge affects letter coding and word identification. In other studies developmental changes in underlying reading processes are inferred from variations across time in the pattern of relationships between different formats of measures of RAN and reading. The overall aim of the symposium is to provide an in depth view on developmental changes in processes between the coding of a letter string and the generation of a correct pronunciation.

Symposium Papers: 

Letter position processing and the development of orthographic knowledge

First Author/Chair:Yvette Kezilas -- Centre of Excellence in Cognition and its Disorders, Department of Cognitive Science, Macquarie University, Australia
Additional authors/chairs: 
Saskia Kohnen ; Meredith McKague ; Serje Robidoux ; Anne Castles

Purpose: Reading and understanding a passage of text requires letters to be assigned to their correct position within a word. Without precise letter position coding, words such as pat would frequently be confused with tap or apt. Studies have shown that the cognitive mechanisms underpinning letter position processing change as reading develops. This study tested the hypothesis that changes in letter position processing are driven specifically by the development of higher-level orthographic knowledge.
Method: We administered a novel variant of the Reicher-Wheeler task to children aged 7-12 years (Experiment 1) and adults (Experiment 2). Participants were asked to report the letter presented at a specified position within three orthographic contexts: anagram words (e.g. form–which has the anagram partner, from), pseudowords (e.g., pilf–plif) and illegal nonwords (e.g. ftkl–fktl). This design enabled us to explore the influence of two components of orthographic knowledge on letter position processing: (1) a reader’s stored whole-word orthographic representations for specific words (measured by the advantage for words over pseudowords; the word superiority effect), and (2) a reader’s awareness of orthotactic constraints (measured by the advantage for pseudowords over illegal nonwords; the pseudoword superiority effect).
Results: The pseudoword superiority effect was found to increase with development in primary school children (Experiment 1). The word superiority effect was only present for adult participants (Experiment 2).
Conclusions: While knowledge of orthotactic constraints assists in letter position processing in primary school children, a reader’s stored whole-word orthographic representations do not influence letter position processing until adulthood.

How morphological knowledge can affect the process of word decoding

First Author/Chair:Carsten Elbro -- University of Copenhagen, Denmark
Additional authors/chairs: 
Megan King; Delisha Rown; Jane Oakhill

Purpose: The present study was concerned with morpheme recognition as a compensatory reading strategy in dyslexic university students. The study compared the degree to which dyslexic and non-dyslexic students utilise morpheme recognition in word decoding. It also investigated the roots of such compensatory mechanisms – and the uses of morphological knowledge in word recognition and reading comprehension. It was hypothesised that dyslexic students might use morphological knowledge both to compensate for phonological weaknesses in decoding – and to boost vocabulary knowledge and reading comprehension directly.
Method: The study employed a combined group comparison and correlational design with 31 dyslexic and 11 non-dyslexic university students. They were individually tested with measures of morpho-syntactic knowledge, morphological word decoding, reading comprehension, and written word identification.
Results: The dyslexic students were outperformed by the non-dyslexic on all measures. However, the dyslexic students utilized a morphological decoding strategy to a significantly higher degree than the non-dyslexic students did. Within the dyslexic group, significant relationships were found between morpho-syntactic knowledge, word identification and reading comprehension. These links were fully mediated by morpheme recognition in word decoding.
Conclusions: Dyslexic university students appear to compensate for basic phonological difficulties in decoding by relying on a morphological recognition strategy. However, they may not take full advantage of their morphological awareness for a direct support of vocabulary knowledge and comprehension in reading.

Lexical competition and learning to read new words

First Author/Chair:Kate Nation -- University of Oxford, UK

Purpose: We addressed how newly-learned words compete with existing words to discover the extent to which new words become integrated and linked with other words in long-term memory. We focused on whether exposure to words with meaning vs. without meaning influences the ease with which an unfamiliar word becomes familiar, as indexed by whether and when it begins to compete with other words in lexical processing.
Method: We used masked priming with lexical decision. If a prime is unfamiliar, it should behave like a nonword, i.e., produce facilitation for familiar target words that overlap in form. If training is successful in bringing about word learning, the newly-learned word should then behave like a word in masked priming, i.e. inhibit the processing of target words that overlap in form. We manipulated whether providing information about the meaning of the unfamiliar word during exposure influenced learning.
Results: Learning was successful in that exposure to words brought about a change in the pattern of priming. Before exposure, unfamiliar (real) words behaved like nonwords. After exposure however, they behaved like words, competing with existing words and slowing response times. On-going analyses are exploring the effects of meaning on the time course of priming.
Conclusions: Priming can be used to chart the transition from a letter string being unfamiliar to it becoming a familiar word. This provides a sensitive and continuous measure of learning. On-going experiments are using the paradigm to measure how various factors (e.g., frequency, linguistic context) influence orthographic learning.

Tracking reading processes with serial and discrete rapid naming

First Author/Chair:Peter de Jong -- University of Amsterdam, The Netherlands
Additional authors/chairs: 
Madelon van den Boer

Purpose: We have argued that the pattern of relations of serial and discrete RAN with serial and discrete reading can reveal underlying reading processes. Cross sectional studies have suggested that during reading development these relations become format specific. Especially, the stronger relation of discrete reading with discrete RAN than with serial RAN, is informative. Discrete RAN seems a measure of the generation of a phonological code from print. Consequently, a strong relation with discrete reading is suggestive of the parallel generation of a pronunciation from print. In this two wave longitudinal study, we examined changes in the RAN-reading relationship from second through third grade.
Method: Participants were 141 children. Rate measures of serial and discrete word, nonword, letter and digit naming were administered. Words and nonwords were mono-syllabic and had 4 letters.
Results: As expected, at each wave serial reading was more strongly correlated with serial RAN than with discrete RAN, whereas discrete RAN was the stronger correlate of discrete reading. Similarly, over time serial RAN was the stronger correlate of serial reading. Unexpectedly, Grade 2 discrete and serial naming were similarly related to and predicted equal percentages of unique variance in Grade 3 discrete reading.
Conclusion: The format specific relations at each wave between RAN and reading rate suggest that even second grade children generate word and nonword pronunciations largely in parallel. Across time, RAN-reading relations were only partly format specific. This finding will be discussed with regards to the stability of discrete measures of RAN and reading.

Serial superiority in rapid naming as a model of fluency development

First Author/Chair:Athanassios Protopapas -- University of Athens, Greece
Additional authors/chairs: 
Katerina Katopodi; Angeliki Altani; George K. Georgiou

Purpose: To study the benefit from simultaneous presentation of stimuli in rapid naming tasks. We hypothesized that word naming initially permits no overlapped processing of consecutive stimuli but eventually develops into an automatized task involving highly overlapping (cascaded) processing.
Method: About 100 Greek children from each of Grades 1, 3, 5 were tested in discrete and serial naming of digits, number words, dice, words, and objects. The same four numbers were used in the first three tasks, matched in structure and frequency to the words and objects, permitting direct comparability of naming times. Durations of recorded responses were submitted to SEM including discrete and serial factors plus an additional factor. Serial superiority was defined as the benefit (in seconds per item) of serial vs. corresponding discrete tasks.
Results: A “reading” factor emerged in early grades, encompassing word tasks, which shared little variance with other serial tasks in Grade 1. In contrast, a “semantic” factor emerged in Grade 5, including object and dice tasks, which fell behind in serial processing. Serial superiority was highest for digits, even in Grade 1, and reached similar values for all nonsemantic tasks, including words in Grades 3 and 5.
Conclusions: Words start off as special, difficult-to-process objects, but gradually become as easy as closed-set items, achieving high serial superiority consistent with unitization of familiar items. The large increase in serial superiority indicates qualitative differences in processing multiword sequences, underlying fluency development beyond increased efficiency of single-word recognition.