Arabic Handwritten Word Recognition Based on Stationary Wavelet Transform Technique using Machine Learning

This paper is aimed at improving the performance of the word recognition system (WRS) of handwritten Arabic text by extracting features in the frequency domain using the Stationary Wavelet Transform (SWT) method using machine learning, which is a wavelet transform approach created to compensate for the absence of translation invariance in the Discrete Wavelets Transform (DWT) method. The proposed SWT-WRS of Arabic handwritten text consists of three main processes: word normalization, feature extraction based on SWT, and recognition. The proposed SWT-WRS based on the SWT method is evaluated on the IFN/ENIT database applying the Gaussian, linear, and polynomial support vector machine, the k-nearest neighbors, and ANN classifiers. ANN performance was assessed by applying the Bayesian Regularization (BR) and Levenberg-Marquardt (LM) training methods. Numerous wavelet transform (WT) families are applied, and the results prove that level 19 of the Daubechies family is the best WT family for the proposed SWT-WRS. The results also confirm the effectiveness of the proposed SWT-WRS in improving the performance of handwritten Arabic word recognition using machine learning. Therefore, the suggested SWT-WRS overcomes the lack of translation invariance in the DWT method by eliminating the up-and-down samplers from the proposed machine learning method.

Word- and Text-Level Processes Contributing to Fluent Reading of Word Lists and Sentences

In this study, we investigated how word- and text-level processes contribute to different types of reading fluency measures. We aimed to increase our understanding of the underlying processes necessary for fluent reading. The sample included 73 Dutch Grade 3 children, who were assessed on serial word reading rate (familiar words), word-list reading fluency (increasingly difficult words), and sentence reading fluency. Word-level processes were individual word recognition speed (discrete word reading) and sequential processing efficiency (serial digit naming). Text-level processes were receptive vocabulary and syntactic skills. The results showed that word- and text-level processes combined accounted for a comparable amount of variance in all fluency outcomes. Both word-level processes were moderate predictors of all fluency outcomes. However, vocabulary only moderately predicted sentence reading fluency, and syntactic skills merely contributed to sentence reading fluency indirectly through vocabulary. The findings indicate that sequential processing efficiency has a crucial role in reading fluency across various measures besides individual word recognition speed. Additionally, text-level processes come into play when complexity and context availability of fluency measures increases, but the exact timing requires further study. Findings are discussed in terms of future directions and their possible value for diagnostic assessment and intervention of reading difficulties.

The Roles of Consonant, Rime, and Tone in Mandarin Spoken Word Recognition: An Eye-Tracking Study

This study investigated the relative role of sub-syllabic components (initial consonant, rime, and tone) in spoken word recognition of Mandarin Chinese using an eye-tracking experiment with a visual world paradigm. Native Mandarin speakers (all born and grew up in Beijing) were presented with four pictures and an auditory stimulus. They were required to click the picture according to the sound stimulus they heard, and their eye movements were tracked during this process. For a target word (e.g., tang2 “candy”), nine conditions of competitors were constructed in terms of the amount of their phonological overlap with the target: consonant competitor (e.g., ti1 “ladder”), rime competitor (e.g., lang4 “wave”), tone competitor (e.g., niu2 “cow”), consonant plus rime competitor (e.g., tang1”soup”), consonant plus tone competitor (e.g., tou2 “head”), rime plus tone competitor (e.g., yang2 “sheep”), cohort competitor (e.g., ta3 “tower”), cohort plus tone competitor (e.g., tao2 “peach”), and baseline competitor (e.g., xue3 “snow”). A growth curve analysis was conducted with the fixation to competitors, targets, and distractors, and the results showed that (1) competitors with consonant or rime overlap can be adequately activated, while tone overlap plays a weaker role since additional tonal information can strengthen the competitive effect only when it was added to a candidate that already bears much phonological similarity with the target. (2) Mandarin words are processed in an incremental way in the time course of word recognition since different partially overlapping competitors could be activated immediately; (3) like the pattern found in English, both cohort and rime competitors were activated to compete for lexical activation, but these two competitors were not temporally distinctive and mainly differed in the size of their competitive effects. Generally, the gradation of activation based on the phonological similarity between target and candidates found in this study was in line with the continuous mapping models and may reflect a strategy of native speakers shaped by the informative characteristics of the interaction among different sub-syllabic components.

Interactions of stimulus quality and semantic context on N400 in visual word recognition

The joint effects of stimulus quality and semantic context in visual word recognition were examined with event-related potential (ERP) recordings. In one-character Chinese word recognition, we manipulated stimulus quality at two degradation levels (highly vs. slightly degraded) and semantic context at two priming levels (semantically related vs. unrelated). In a prime–target–probe trial flow, ERPs were recorded to the target character which was presented in either high or slight degradation and which was preceded by either a semantically related or unrelated prime character. The target character was then followed by a probe character which was either identical to or different from the target character. Subjects were instructed to make target–probe matching judgments. The ERP results demonstrated a degradation by priming interaction, with larger N400 semantic priming effects for slightly degraded targets. Moreover, the degradation effects were observed on the P200, N250, and N400. These findings provided evidence for the cascaded model of visual word recognition such that the visual processing cascaded into the semantic stage and thus interacted on the N400 amplitude. The results were compared to an earlier study with a null ERP degradation by priming interaction. The ramifications of these results for models of visual word recognition are discussed.

Addressing Word Recognition Difficulties of Adolescents With Dyslexia

This chapter provides a comprehensive overview of evidence-based word analysis approaches for promoting accurate and fluent reading of complex words by adolescents with a specific reading disability (i.e., dyslexia). First, research has been reviewed to pinpoint the characteristics and causes of dyslexia as a specific learning disability. Specifically, two theories of dyslexia, the phonological theory of dyslexia and the magnocellular theory of dyslexia, have been discussed to ascertain the causal attributes of phonological awareness deficits and auditory and visual sequencing deficits to word recognition difficulties of adolescents with dyslexia. Next, two theories of word recognition, particularly the dual-route model of word recognition and connectionist model of word recognition, have been discussed to clarify the mechanism underlying the manifestation of dyslexia and resultant difficulties with word recognition. Finally, evidence-based word analysis programs have been described as approaches for improving word reading ability of adolescents with dyslexia.