Search

Simple numerical versions of the Spatial Coding and of the Open Bigrams coding of character strings are presented, together with a natural merging of these two approaches. Comparing the predictive performance of these three orthographic coding schemes on orthographic masked priming data, we observe that the merged coding scheme always provides the best fits. Testing the ability of the orthographic codes, used as regressors, to capture relevant regularities in lexical decision data, we also observe that the merged code provides the best fits and that both the spatial coding component and the open bigrams component provide specific and significant contributions. This gives us a new lighting on probable mechanisms involved in orthographic coding, together with new tools for modelling behavioural and electrophysiological data collected in word recognition tasks.

Many neural learning algorithms require to solve large least square systems in order to obtain synaptic weights. Moore-Penrose inverse matrices allow for solving such systems, even with rank deficiency, and they provide minimum-norm vectors of synaptic weights, which contribute to the regularization of the input-output mapping. It is thus of interest to develop fast and accurate algorithms for computing Moore-Penrose inverse matrices. In this paper, an algorithm based on a full rank Cholesky factorization is proposed. The resulting pseudoinverse matrices are similar to those provided by other algorithms. However the computation time is substantially shorter, particularly for large systems., Comment: Number of pages: 5. Typo page 26 line 3: one must read W=G^+F (instead of W=G^+W, which does not make sense!)

Unpublished manuscript, 14 pages.; This paper describes two experiments using a masked priming method with 60 ms SOA. In the first experiment, the task was an alphabetical decision. The stimuli were isolated letters or non-alphabetical symbols, preceded by a similar or different prime, while the primes were scaled down or 180° rotated. Response times to letters revealed priming effects for both prime transformations. In the second experiment, the task was a lexical decision, and the stimuli were five lower-case letter words or pseudo-words. The priming conditions were similar to those of the first experiment. Response times to words revealed priming effects for both prime transformations, however the priming effect was only marginally significant for rotated primes and it appeared dependent on the frequency of use of the prime. A significant correlation between priming effects and the frequency of use of the different prime words was observed. We concluded that scale invariant features are used in the perception of letters and words as well, while 180° rotation invariant features are used in the perception of letters, but no such a conclusion can be drawn for words, in general.

Courrieu Pierre. Analyse-t-on le mot de gauche à droite ?. In: Bulletin de psychologie, tome 39 n°375, 1986. Jugement et langage. Hommage à Georges Noizet. pp. 425-427.

SUMMARYDocument referenceCourrieu, P. (1983). L'identification des mots au cours de la lecture [Word recognition in reading]. Doctoral thesis, Vol. 1, Aix-Marseille 1 University, Aix-en-Provence, December 1983, 409 p.OriginThis document is a doctoral thesis in French, defended on December 22, 1983, at Aix-Marseille 1 University. The original paper document is available, together with two annex volumes including experimental materials and raw data, at Aix-Marseille University library (Aix-Schuman), cote TLD 3351/1*BULA (Vol. 1), TLD 3351/2*BULA (Vol. 2), and TLD 3351/3*BULA (Vol. 3).ContentsThis is a series of 12 experimental studies on the perceptual encoding processes involved in reading printed words silently. In a preliminary experiment (pp. 97-107), one exploited the fact that all nouns in French language have a gender. So, one can build grammatical nominal syntagms of the form LE PHARE, non-grammatical syntagms of the form LE NEIGE, and non-grammatical syntagms that are homophonic of grammatical syntagms, such as LE PAIRE, which is a homophone of the correct syntagm LE PERE. It was observed that when participants had to judge the grammaticality of such syntagms in an orthographic judgment task, it was more difficult to reject homophone syntagms than to reject non-homophone non-grammatical syntagms. When the participants were requested to make their judgment uniquely based on the phonological form of the stimuli, it was more difficult to accept homophone syntagms than to accept orthographically grammatical syntagms. It was concluded that both the use of an orthographic encoding and of a phonological encoding were automatic and irrepressible in printed word recognition. The next 11 experiments concerned orthographic coding. In experiment 1, fast left-to-right scanning effects involved in the discrimination of words were observed in a lexical decision task, while in experiment 2, left-to-right scanning effects were observed in an alphabetic decision task on rapidly presented non-words. Experiments 3-11 all used masked priming type techniques with a lexical decision task. Experiments 3 and 4 showed that non-alphabetical primes produce less masking effect than alphabetical primes, and that repeated letter primes produce less masking effect than primes made of different letters. This suggests that different occurrences of the same letter are not processed independently. Experiment 5 showed that the processing of letters has a local component and also a non-local one. The results also suggest that left-to-right serial effects observed in this experiment are not related to phonological encoding. Experiments 5-11 tested the effects of primes that were anagram of the target with various structures. It was concluded that non-positional letter codes activate the lexicon, and that bigrams are used to verify the structure of the stimulus, while the initial bigram is, in fact, made of the left space followed by the first letter of the word, and it plays a very important role in the word recognition. It was also observed that reverse bigrams (e.g. ON / NO) are perceptually antagonist, the first presented bigram tending to mask the second one, no matter their respective positions. A simple neuron-like model is proposed that detects bigrams from sequential responses of non-positional letter detectors. This model spontaneously generates the reverse bigram effect and makes new predictions concerning the perception of palindromes (RADAR).; Auteur: Courrieu, PierreTitre: L'identification des mots au cours de la lecture. Document: Thèse de doctorat de 3ème cycle en Psychologie (Vol. 1), Université Aix-Marseille 1, Aix-en-Provence, 22 December 1983, 409 p.Disponibilité: Le document papier original est disponible, ainsi que deux volumes annexes contenant les matériels expérimentaux, les analyses statistiques, et les données brutes, à la bibliothèque de l'Université Aix-Marseille (AMU, Aix-Schuman), cote TLD 3351/1*BULA (Vol. 1), TLD 3351/2*BULA (Vol. 2), et TLD 3351/3*BULA (Vol. 3).