Bernard Lété

MANULEX: A grade-level lexical database from French elementary school readers

This article presents MANULEX, a Web-accessible database that provides grade-level word frequency lists of nonlemmatized and lemmatized words (48,886 and 23,812 entries, respectively) computed from the 1.9 million words taken from 54 French elementary school readers. Word frequencies are provided for four levels: first grade (G1), second grade (G2), third to fifth grades (G3-5), and all grades (G1-5). The frequencies were computed following the methods describedby Carroll, Davies, and Richman (1971) and Zeno, Ivenz, Millard, and Duwuri (1995), with four statistics at each level (F, overall word frequency;D, index of dispersion across the selectedreaders;U, estimated frequencyper million words; andSFI, standard frequency index). The database also provides the number of letters in the word and syntactic category information. MANULEX is intended to be a useful tool for studying language development through the selection of stimuli based on precise frequency norms. Researchers in artificial intelligence can also use it as a source of information on natural language processing to simulate written language acquisition in children. Finally, it may serve an educational purpose by providing basic vocabulary lists.

Manulex-infra: Distributional characteristics of grapheme-phoneme mappings, and infralexical and lexical units in child-directed written material

It is well known that the statistical characteristics of a language, such as word frequency or the consistency of the relationships between orthography and phonology, influence literacy acquisition. Accordingly, linguistic databases play a central role by compiling quantitative and objective estimates about the principal variables that affect reading and writing acquisition. We describe a new set of Web-accessible databases of French orthography whose main characteristic is that they are based on frequency analyses of words occurring in reading books used in the elementary school grades. Quantitative estimates were made for several infralexical variables (syllable, grapheme-to-phoneme mappings, bigrams) and lexical variables (lexical neighborhood, homophony and homography). These analyses should permit quantitative descriptions of the written language in beginning readers, the manipulation and control of variables based on objective data in empirical studies, and the development of instructional methods in keeping with the distributional characteristics of the orthography.

Evidence for multiple routes in learning to read

We describe a multiple-route model of reading development in which coarse-grained orthographic processing plays a key role in optimizing access to semantics via whole-word orthographic representations. This forms part of the direct orthographic route that gradually replaces phonological recoding during the initial phases of reading acquisition. The model predicts distinct developmental trajectories for pseudo-homophone and transposed-letter effects - two benchmark phenomena associated with phonological recoding and coarse-grained orthographic processing, respectively. Pseudo-homophone effects should decrease over the first years of reading acquisition, whereas transposed-letter effects should initially increase. These predictions were tested in a lexical decision task with 334 children in grades 1-5, and 29 skilled adult readers. In line with the predictions, we found that the pseudo-homophone effect diminished as reading level increased, whereas the transposed-letter effect first increased and then diminished.

Orthographic and Phonological Contributions to Reading Development: Tracking Developmental Trajectories Using Masked Priming

The present study used a variant of masked priming to track the development of 2 marker effects of orthographic and phonological processing from Grade 1 through Grade 5 in a cross-sectional study. Pseudohomophone (PsH) priming served as a marker for phonological processing, whereas transposed-letter (TL) priming was a marker for coarse-grained orthographic processing. The results revealed a clear developmental picture. First, the PsH priming effect was significant and remained stable across development, suggesting that phonology not only plays an important role in early reading development but continues to exert a robust influence throughout reading development. This finding challenges the view that more advanced readers should rely less on phonological information than younger readers. Second, the TL priming effect increased monotonically with grade level and reading age, which suggests greater reliance on coarse-grained orthographic coding as children become better readers. Thus, TL priming effects seem to be a good marker effect for childrens ability to use coarse-grained orthographic coding to speed up direct lexical access in alphabetic languages. The results were predicted by the dual-route model of orthographic processing, which suggests that direct orthographic access is achieved through coarse-grained orthographic coding that tolerates some degree of flexibility in letter order.

Learning to spell from reading: general knowledge about spelling patterns influences memory for specific words.

Adults often learn to spell words during the course of reading for meaning, without intending to do so. We used an incidental learning task in order to study this process. Spellings that contained double n, r and t which are common doublets in French, were learned more readily by French university students than spellings that contained less common but still legal doublets. When recalling or recognizing the latter, the students sometimes made transposition errors, doubling a consonant that often doubles in French rather than the consonant that was originally doubled (e.g., tiddunar recalled as tidunnar). The results, found in three experiments using different nonwords and different types of instructions, show that people use general knowledge about the graphotactic patterns of their writing system together with word-specific knowledge to reconstruct spellings that they learn from reading. These processes contribute to failures and successes in memory for spellings, as in other domains.

An Adaptive Resonance Theory account of the implicit learning of orthographic word forms

An Adaptive Resonance Theory (ART) network was trained to identify unique orthographic word forms. Each word input to the model was represented as an unordered set of ordered letter pairs (open bigrams) that implement a flexible prelexical orthographic code. The network learned to map this prelexical orthographic code onto unique word representations (orthographic word forms). The network was trained on a realistic corpus of reading textbooks used in French primary schools. The amount of training was strictly identical to childrens exposure to reading material from grade 1 to grade 5. Network performance was examined at each grade level. Adjustment of the learning and vigilance parameters of the network allowed us to reproduce the developmental growth of word identification performance seen in children. The network exhibited a word frequency effect and was found to be sensitive to the order of presentation of word inputs, particularly with low frequency words. These words were better learned with a randomized presentation order compared with the order of presentation in the school books. These results open up interesting perspectives for the application of ART networks in the study of the dynamics of learning to read.

Word-shape and word-lexical-frequency effects in lexical-decision and naming tasks

The present study investigated the use of word-shape information in visual word recognition. Word-shape frequency was computed for lexically frequent and rare words. Experiment 1 contrasted lowercase and uppercase presentations in a lexical-decision task. The observed latencies indicated a shape-frequency effect in lowercase presentation, i.e., responses were given faster for words with a low-frequency shape than for words with a high-frequency shape, and an interaction between shape frequency and lexical frequency, indicating that, for rare words, having a rare shape speeded up lexical decisions. Experiment 2 primed shape information with a 400 ms SOA. The results showed that high-frequency shape words benefited more from the priming procedure than did low-frequency ones. Priming was also used in a naming task (Experiment 3). The results indicated a strong priming effect for all four target types. When all words were given an up-down configuration (Experiment 4), the same pattern of results as in Experiment 1 was found, rejecting a letter-confusability explanation. Taken together, the results suggest that shape information affects word recognition. Having a rare shape seems to shorten lexical decision times on lexically rare words and to lengthen naming times on lexically frequent words.

Visual and linguistic determinants of the eyes' initial fixation position in reading development☆

Two eye-movement experiments with one hundred and seven first- through fifth-grade children were conducted to examine the effects of visuomotor and linguistic factors on the recognition of words and pseudowords presented in central vision (using a variable-viewing-position technique) and in parafoveal vision (shifted to the left or right of a central fixation point). For all groups of children, we found a strong effect of stimulus location, in both central and parafoveal vision. This effect corresponds to the childrens apparent tendency, for peripherally located targets, to reach a position located halfway between the middle and the left edge of the stimulus (preferred viewing location, PVL), whether saccading to the right or left. For centrally presented targets, refixation probability and lexical-decision time were the lowest near the words center, suggesting an optimal viewing position (OVP). The viewing-position effects found here were modulated (1) by print exposure, both in central and parafoveal vision; and (2) by the intrinsic qualities of the stimulus (lexicality and word frequency) for targets in central vision but not for parafoveally presented targets.

A developmental perspective on visual word recognition: New evidence and a self-organising model

This study investigated the developmental trajectory of two marker effects of visual word recognition, word frequency, and orthographic neighbourhood effects, in French primary school children from Grades 1 to 5. Frequency and neighbourhood size were estimated using a realistic developmental database, which also allowed us to control for the effects of age-of-acquisition. A lexical decision task was used because the focus of this study was orthographic development. The results showed that frequency had clear effects that diminished with development, whereas orthographic neighbourhood had no significant influence at either grade. A self-organising neural network was trained on the realistic developmental corpus. The model successfully simulated the overall pattern found with children, including the absence of neighbourhood size effects. The self-organising neural network outperformed the classic interactive activation model in which frequency effects are simulated in a static way. These results highlight the potentially important role of unsupervised learning for the development of orthographic word forms.

A developmental investigation of the first-letter advantage

Skilled adult readers identify the first letter in a string of random consonants better than letters at any other position, and this advantage for the initial position is not seen with strings of symbols or familiar shapes. Here we examined the developmental trajectory of this first-letter advantage by testing children in Grades 1 to 5 of primary education in a target-in-string identification paradigm. Strings of five letters or five simple shapes were briefly presented, and children were asked to identify a target letter/shape at one of the five possible positions. Children responded by choosing between the target and an alternative that was a neighboring letter/shape (e.g., TPFMR-M vs. F at position 4). The serial position function linking accuracy to position-in-string was found to be affected by reading ability differently for letter stimuli compared with shape stimuli, and this was found to be almost entirely driven by differences in performance in identifying targets at the first position in strings. Here, accuracy increased more rapidly for letter stimuli than for shape stimuli as reading ability increased. This developmental pattern, plus the fact that letter strings were composed of random consonants and the task minimized the involvement of verbal recoding, allows us to exclude an explanation of the first-letter advantage in terms of serial reading strategies or phonological decoding. The findings suggest that the first-letter advantage is a function of, and a marker for, increasingly efficient orthographic processing.