EXPRESS: The letter position coding mechanism of second language words during sentence reading: Evidence from eye movements

We conducted three eye movement experiments to investigate the mechanism for coding letter positions in a person’s second language during sentence reading; we also examined the role of morphology in this process with more rigorous manipulation. Given that readers not only obtain information from currently fixated words (i.e., the foveal area) but also from upcoming words (i.e., the parafoveal area) to guide their reading, we examined both when the targets were fixated (Exp. 1) and when the targets were seen parafoveally (Exp. 2 and Exp. 3). First, we found the classic transposed letter (TL) effect in Exp. 1, but not in Exp. 2 or Exp. 3. This implies that flexible letter position coding exists during sentence reading. However, this was limited to words located in the foveal area, suggesting that L2 readers whose L2 proficiency is not as high as skilled native readers are not able to extract and utilize the parafoveal letter identity and position information of a word, whether the word length is long (Exp. 2) or short (Exp. 3). Second, we found morphological information to influence the magnitude of the TL effect in Exp. 1. These results provide new eye movement evidence for the flexibility of L2 letter position coding during sentence reading, as well as the interactions between the different internal representations of words in this process. Altogether, this is helpful for understanding L2 sentence reading and visual word recognition. Thus, future L2 reading frameworks should integrate word recognition and eye movement control models.

Do Grading Gray Stimuli Help to Encode Letter Position?

Numerous experiments in the past decades recurrently showed that a transposed-letter pseudoword (e.g., JUGDE) is much more wordlike than a replacement-letter control (e.g., JUPTE). Critically, there is an ongoing debate as to whether this effect arises at a perceptual level (e.g., perceptual uncertainty at assigning letter position of an array of visual objects) or at an abstract language-specific level (e.g., via a level of “open bigrams” between the letter and word levels). Here, we designed an experiment to test the limits of perceptual accounts of letter position coding. The stimuli in a lexical decision task were presented either with a homogeneous letter intensity or with a graded gray intensity, which indicated an unambiguous letter order. The pseudowords were either transposed-letter pseudowords or replaced-letter pseudowords (e.g., jugde vs. jupte). The results showed much longer response times and substantially more errors in the transposed-letter pseudowords than in the replacement-letter pseudowords, regardless of visual format. These findings favor the idea that language-specific orthographic element factors play an essential role when encoding letter position during word recognition.

Graph Convolutional Networks with Bidirectional Attention for Aspect-Based Sentiment Classification

Aspect-based sentiment classification aims at determining the corresponding sentiment of a particular aspect. Many sophisticated approaches, such as attention mechanisms and Graph Convolutional Networks, have been widely used to address this challenge. However, most of the previous methods have not well analyzed the role of words and long-distance dependencies, and the interaction between context and aspect terms is not well realized, which greatly limits the effectiveness of the model. In this paper, we propose an effective and novel method using attention mechanism and graph convolutional network (ATGCN). Firstly, we make full use of multi-head attention and point-wise convolution transformation to obtain the hidden state. Secondly, we introduce position coding in the model, and use Graph Convolutional Networks to obtain syntactic information and long-distance dependencies. Finally, the interaction between context and aspect terms is further realized by bidirectional attention. Experiments on three benchmarking collections indicate the effectiveness of ATGCN.

What, Where, When and How of Visual Word Recognition: A Bibliometrics Review

The neural/mental operations involved in the process of visual word recognition (VWR) are fundamental for the efficient comprehension of written/printed words during reading. The present study used CiteSpace, a visual analysis software, to identify the intellectual landscape where VWR has been reviewed in the past decade. Thus, synthesized co-citation networks were analyzed to explore and discuss the main questions raised in the VWR literature: the research fronts and the emerging trends of research on this topic. Our results showed that the main questions addressed in VWR studies during the last decade have been focused on four main aspects related to “what,” “where,” “when,” and “how” of VWR; to be specific, the different types of representations assessed during VWR (“what”), the locations and the timing of the brain activity involved in VWR (“where” and “when”), and the interactivity among different representations during processing (“how”). Among the revised studies, letter position coding was found to be the main topic of interest, possibly reflecting the critical role of this process. Furthermore, the evidence found in these studies consistently supported that VWR implies access to phonological, semantic, and morphological representations, which interact and modulate the processing of written words, particularly during early stages. Altogether, our findings showed the evolution in VWR literature regarding the different cognitive and neural operations involved in this process, highlighting the growing interest over the last decade toward the top-down way that mental representations interact.

On the noisy spatiotopic encoding of word positions during reading: Evidence from the change-detection task

The present study builds on our prior work showing evidence for noisy word-position coding in an immediate same-different matching task. In that research, participants found it harder to judge that two successive brief presentations of five-word sequences were different when the difference was caused by transposing two adjacent words compared with different word replacements – a transposition effect. Here we used the change-detection task with a 1-s delay introduced between sequences – a task thought to tap into visual short-term memory. Concurrent articulation was used to limit the contribution of active rehearsal. We used standard response-time (RT) and error-rate analyses plus signal detection theory (SDT) measures of discriminability (d’) and bias (c). We compared the transposition effects for ungrammatical word sequences and nonword sequences observed with these different measures. Although there was some evidence for transposition effects with nonwords, the effects were much larger with word sequences. These findings provide further support for the hypothesized noisy assignment of word identities to spatiotopic locations along a line of text during reading.

Morpheme position coding in reading development as explored with a letter search task

Suffixes have been shown to be recognized as units of processing in visual word recognition and their identification has been argued to be position-specific in skilled adult readers: in lexical decision tasks suffixes are automatically identified at word endings, but not at word beginnings. The present study set out to investigate whether position-specific coding can be detected with a letter search task and whether children already code suffixes as position-specific units. A preregistered experiment was conducted in Italian in which 3rd-graders, 5th-graders, and adults had to detect a target letter that was either contained in the suffix of a pseudoword (e.g., S in flagish vs. flagosh) or in a non-suffix control (e.g., S in flagish vs. flagosh). To investigate sensitivity to position, letters also had to be detected in suffixes and non-suffixes placed in reversed position, that is in the beginning of pseudowords (e.g., S in ishflag vs. oshflag). Results suggested position-specific processing differences between suffixes and non-suffixes that develop throughout reading development. However, some effects were weak and only partially compatible with the hypotheses. Therefore, a second experiment was conducted. The effects of position-specific suffix identification could not be replicated. A combined analysis additionally using a Bayesian approach indicated no processing differences between suffixes and non-suffixes in our task. We discuss potential interpretations and the possibility of letter search being unsuited to investigate small processing differences. We connect our example of failed self-replication to the current discussion about the replication crisis in psychology and the lesson psycholinguistics can learn.