Cette these s'inscrit dans le domaine de la psychologie cognitive du langage et porte sur le role des connaissances phonologiques et semantiques dans le traitement lexical. Nous nous proposons d'examiner l'interaction entre ces variables dans le comportement des lecteurs adultes de langue maternelle francaise. La premiere etude de cette these porte sur l'etablissement de normes de frequence subjective et d'imagerie mentale pour 1,760 mots monosyllabiques de la langue francaise. Cette etude permet l'etablissement de normes stables et, en complement aux normes deja publiees par Desrochers et Bergeron (2000), fournit des normes pour la quasi totalite des mots monosyllabiques francais. La deuxieme etude porte sur la lecture a voix haute et sur l'impact de la valeur d'imagerie lors du traitement des mots orthographiquement consistants (i.e. pomme) et inconsistants (i.e. clef). Cette etude illustre le role important de la semantique dans une tache de nature phonologique par la demonstration d'un effet de la valeur d'imagerie sur les mots inconsistants de frequence forte autant que faible. Enfin, la troisieme etude traite de l'impact de la phonologie sur le traitement semantique dans une tache de discrimination semantique. Les resultats de cette etude montrent que lorsque l'on varie la valeur d'imagerie d'un homophone (distracteur), cette manipulation tend a faciliter les reponses aux mots cibles (bonne reponse), et plus particulierement aux cibles a faible valeur d'imagerie lorsque ces deux composants sont presentes a l'interieur d'un meme essai. Lorsqu'une decision semantique est requise uniquement sur l'homophone, on observe que la valeur d'imagerie n'a pas d'impact sur ce type de mot. En fait, l'effet de la valeur d'imagerie n'importe que pour le traitement de la cible. En resume, la presente recherche permet d'apporter des precisions importantes sur l'interaction entre la phonologie et la semantique, leur generalisation inter-linguistiques ainsi que sur les implications de cette interaction pour les modeles theoriques de la lecture qui reposent sur des postulats architecturaux differents (p.ex., les modeles connexionnistes et les modeles a deux voies d'acces).

This dissertation investigates the cognitive relevance of selected morphological and phonological units in the Arabic mental lexicon. The morphological units are sound and weak roots, etymons, phonetic matrices, and sound and weak patterns. The phonological units are vowels and consonants. The work is motivated by a controversy in Arabic morphology that is paralleled by a cross-linguistic debate in lexical processing. There are two views in Arabic morphology, the stem-based theory and the morpheme-based theory that is represented by two sub-theories. The first sub-theory argues that derivations are based on roots and patterns and the second proposes that the root should be replaced by the etymon and the phonetic matrix. The morpheme-based theory is congruent with lexical processing hypotheses that propose that complex words are accessed and represented as morphemes. The stem-based theory maintains that derivation is stem or word-based and is in line with the whole word hypothesis of lexical processing. These theoretical positions on Arabic morphology and lexical processing were tested in six priming experiments. One objective of these experiments was to test which of these morphemes prime word recognition. Another objective was to test the prediction of connectionism, another lexical processing hypothesis, that priming time correlates with prime-target overlap. A third objective was to examine how abstract the processing of these morphemes could be. The cognitive status of vowels and consonants was tested using a letter-circling task. The results of the online studies have shown that both roots and etymons facilitate word recognition significantly more than orthographic controls. However, non-ordered etymons, phonetic matrices, and patterns did not facilitate word recognition. Weak roots had priming effects only when primes and targets shared a vague semantic relationship. There was no correlation between priming time and meaning and/or form overlap. The lack of priming with non-ordered etymons suggests that there could be limits on abstractness in lexical processing. The results of the offline task suggest that root consonants are more salient than other letters. On the whole, the results support a morpheme-based theory of Arabic morphology and a localist view of lexical processing that assumes a morphemic stage in word recognition.

Glucose regulation, in both patients with diabetes and nondiabetes patients, has been shown to be associated with neuropsychological function, particularly episodic memory. The main purpose of this study was to examine this association in a large sample of young nondiabetic adults, using a repeated-measures design in which participants were assessed under both, glucose and saccharin conditions. Regression analyses revealed that glucoregulatory indices based on evoked glucose levels significantly predicted the verbal memory performance of 135 young adults, independent of demographic and vascular risk factors. Repeated multiple analyses of variances using cognitive measures by function as dependent variables revealed that worse glucose regulators showed poorer verbal memory performance compared to better glucose regulators, with the performance of average glucose regulators falling in between. There was no effect of solution nor did gender interact with glucose regulation. Although worse regulators showed evidence of hyperinsulinemia, glucoregulatory indices calculated on the basis of insulin levels or fasting glucose levels were less sensitive to cognitive variability relative to indices based on evoked glucose levels. Cardiovascular risk factors were indeed associated with hyperinsulinemia, however, these did not predict cognitive performance in this young group. These results were discussed in light of medial temporal lobe functions, including encoding and retrieval of information, and appear to be mediated by changes in lipid metabolism, insulin receptor distribution and function, and/or genetic susceptibility.

Abstract not available.

Animal studies have shown that there is a dose-response impact of glucose on memory but very few studies in humans have used several doses of glucose adjusted to body weight to determine whether there is more than one optimal dose. This study explored further the effect of ingested glucose on cognition in older adults by examining the impact of three doses of glucose (300 mg/kg, 650 mg/kg and 1 g/kg) on performance on neuropsychological tasks, including cognitive measures involving a concomitant or interference task. Another goal of this study was to investigate whether glucoregulation will mediate the impact of glucose on cognitive performance and if age will interact with this effect. A total of 103 older adults (mean age = 68.22) were tested twice, once after glucose ingestion and once after saccharin. Participants were categorized as younger (60-67) and older (68-85) and as better or worse glucoregulators. A doubly-multivariate analysis was conducted on 19 cognitive measures and only showed an overall significant effect of age. Some results suggest a small but not significant overall general decrease in performance on many tasks in people with worse glucoregulation. A number of methodological issues such as age and the generally high level of functioning of the participants may have reduced the ability of the present study to observe cognitive deficits or improvements.

The focused sense of self that is central to our experience of consciousness could not exist without human brain's ability to process large amounts of information outside of conscious awareness. The Mismatch Negativity (MMN) is thought to index one such pre-conscious mechanism, a complex novelty detector that compares incoming sounds to rules that it has extracted from the recent acoustic past. Processes that do not require conscious awareness are considered to be "automatic". The automaticity of the MMN, however, has predominantly been studied using very simple stimuli. The present dissertation studied the issue of automaticity using a two-tone alternating pattern (e.g., ABABABAB...) as the "standard" stimuli. MMN-eliciting deviants were rule-violating tone repetitions (e.g., ABABAA&barbelow;AB...). Three studies took complementary approaches to the problem of automaticity. Study 1 tested the automaticity of the detection of rule violations by varying the presumed attentional demands of a visual task, while the auditory pattern was ignored. The MMN was unaffected by visual task difficulty, but it may be impossible for awake, alert subjects to ever fully ignore incoming sounds. Study 2, therefore, studied the MMN during natural sleep, the period of time when the observer is least conscious of the external environment. An MMN was elicited in both the waking state and REM sleep. Interpreted in isolation, these studies might be taken as evidence that the MMN is strongly automatic. Some attention effects, however, have been reported. One proposed mechanism for these attention effects is that attention increases the strength of the memory for the frequently-presented, standard stimuli. Study 3 examined whether the MMN varies with the strength of the memory for the standard by manipulating the number of memory-reinforcing repetitions of the standard that occurred between successive deviants. The results of Study 3 suggest that the amplitude of the MMN is unrelated to the strength of the memory for the standard. Altogether, the results of these studies suggest that even complex, rule-based MMN elicitation is an automatic process, and that studies demonstrating evidence to the contrary should be examined for the influence of other confounding factors.

Existing evidence on the role of the syllable in visual word recognition is insufficient to inform the extention of models of single-syllable word reading to the more general case of multi-syllable words. Open questions include what syllable units are relevant (e.g., Max Onset, Max Coda), when and why are they relevant, and for whom? A common paradigm for addressing this issue is the word-splitting manipulation in visual lexical decision, where a visual boundary is introduced within words; this boundary is either consistent or inconsistent with that specified by a theoretical syllable unit. If an advantage is observed for the consistent condition, then a 'syllable effect' has been obtained. Recent work has focused on identifying variables that moderate the effect of such word-splitting manipulations in an attempt to explain the empirical inconsistency in this literature (Chen & Vaid, 2007; Taft, 2001, 2002). This line of enquiry was pursued in two studies employing the word-splitting paradigm in lexical decision. The first study reports a series of three experiments (N=48 x 3) where a combination of random coefficient analysis and multi-level modeling was used to examine the joint contribution of syllabic complexity, lexical frequency, and the tendency of participants to rely on phonology. The results indicate that participants who are most sensitive to homophone interference tend to prefer the 'phonological' Max Onset syllable with low-frequency words, situating the locus of this effect in lexical phonology. A follow-up study (N=122) using the same analytical technique with a broad set of indicators provides some corroboration of this effect, and additional information suggesting that apparent patterns of syllable preference arise from multiple causes, some of which depend on Print Exposure and/or participant response speed. The results highlight the need to disentangle response speed from other participant characteristics in predicting syllable preference.

The importance of continuous sight of the target in 7.5 month old infants' reaching was explored in a task that addressed the issues of infants' ability to anticipate and to retain information about the properties of a hidden object. Barriers and darkness were used to investigate infants' ability to compensate for the physical and visual obstruction of a target object in a reaching task. Infants' ability to intercept a moving object with a partially obscured trajectory was tested. Thirty 7.5 month old infants were presented with a graspable object that moved in a straight-line path through their reaching space. In some conditions the object was obscured by a barrier or by darkness for one second just prior to moving within reach, and infants' frequency of reaching and success at contacting the object were used to evaluate their performance. Further analyses of the infants' looking behavior and of the path of their reaching hand helped to clarify the reasons underlying their successes and failures. Infants showed some ability to adapt to a loss of visual information about the moving target object's position by sometimes successfully contacting the object in the barrier conditions. However, infants reached less often and with less success when access to, or sight of, the target object was obstructed. The infants' visual tracking, obstacle-avoidance skills, and ability to retain information about a hidden object were examined in conjunction with kinematic data to explain infants' limitations in adapting to obstacles in reaching tasks. These limitations involved difficulty visually tracking the object past a barrier, particularly in the dark conditions, as well as difficulty successfully aiming a reach around a barrier. Infants appeared to ignore the path of their hand on its way toward the target object, resulting in the hand frequently contacting a barrier rather than the target. Infants' successful contacts in the barrier conditions suggest that they do not require constant visual information about target position in order to enact a proficient reach. Further, infants appear to predict the reappearance of the target object and remember the path and speed of the object during its occlusion.

The visual processing of a stimulus is facilitated by attention when it is at an attended location compared to an unattended location. However, whether attentional selection operates on the basis of visual features (e.g., color) independently of spatial locations is less clear. Six experiments were designed to examine how color information as well as location information affected attentional selection. In Experiment 1, the color of the targets and the spatial distance between them were both manipulated. Stimuli were found to be grouped based on color similarity. Additionally, the evidence suggested direct selection on the basis of color groups, rather than selection that was mediated by location. By varying the probabilities of target location and color, Experiments 2, 3 and 4 demonstrated that the use of color in perceptual grouping and in biasing the priority of selection is not automatic, but is modulated by task demands. Experiments 5 and 6 further investigated the relationship between using color and using location as the selection basis under exogenous and endogenous orienting. The results suggest that the precise nature of the interaction between color and location varies according to the mode of attentional control. Collectively, these experiments contribute to an understanding of how different types of information are used in selection and suggest a greater degree of flexibility of attentional control than previously expected. The flexibility is likely to be determined by a number of factors, including task demands and the nature of attentional control.

In this dissertation, two exemplar-based models of categorization, the General Context Model (GCM) and the Exemplar Based Random Walk model (EBRW), were used to describe between-group categorization differences in artificial and natural language categories. Prior research has shown that political Conservatives in avoidance mode are more exclusive categorizers of natural language category members than Conservatives in approach mode, but this effect was absent for Liberals (Rock & Janoff-Bulman, 2010). In Experiment 1, experimenter-generated stimuli were used to show that the EBRW could account for between-group differences in categorization decisions. In Experiment 2, the data collected by Rock and Janoff-Bulman were used to develop techniques allowing the GCM to account for between-group differences in natural language categorization decisions. Experiment 3 extends these methods to allow the EBRW to account for between-group differences in natural language categorization decisions. Across these experiments, the models identify between-group differences in determining similarity, bias to give an "in-the-category" decision, and the amount of information required to make a categorization decision. Techniques for modeling natural language categorization decisions are discussed.