Explore the vast range of titles available.

Detection of word boundaries in continuous speech is a tedious process due to the absence of a definite pause or silence in the word boundary position. Thus, continuous speech recognition is a very challenging task. However, the prosodic word boundaries, unlike the written word boundaries, can be predicted using the prosodic parameters of continuous speech. This paper proposes a method for detecting such prosodic word boundaries from Bengali continuous speech. Bengali is a bound-stress language, where stress is observed on the first syllable of a prosodic word. Empirical Mode Decomposition is applied to the logarithm of fundamental frequency (F 0 ) contour of continuous speech to detect prosodic word boundaries. 200 Bengali readout sentences, read by ten speakers, are analyzed for the present work. An overall prosodic boundary detection accuracy of 88.05% is achieved, whereas precision and recall values are 90.73% and 88.31%, respectively, with f-score as 89.5. A prosodic word dictionary comprising 5031 prosodic words has been developed by analyzing 1526 Bengali sentences with the proposed prosodic word boundary detection method.

Unlike English and other Western languages, many Asian languages such as Chinese and Japanese do not delimit words by space. Word segmentation and new word detection are therefore key steps in processing these languages. Chinese word segmentation can be considered as a part-of-speech (POS)-tagging problem. We can segment corpus by assigning a label for each character which indicates the position of the character in a word (e.g., “B” for word beginning, and “E” for the end of the word, etc.). Chinese word segmentation seems to be well studied. Machine learning models such as conditional random field (CRF) and bi-directional long short-term memory (LSTM) have shown outstanding performances on this task. However, the segmentation accuracies drop significantly when applying the same approaches to out-domain cases, in which high-quality in-domain training data are not available. An example of out-domain applications is the new word detection in Chinese microblogs for which the availability of high-quality corpus is limited. In this paper, we focus on out-domain Chinese new word detection. We first design a new method Edge Likelihood (EL) for Chinese word boundary detection. Then we propose a domain-independent Chinese new word detector (DICND); each Chinese character is represented as a low-dimensional vector in the proposed framework, and segmentation-related features of the character are used as the values in the vector.

Chinese words often exhibit a parallel structural relationship within sentences, while individual characters are sequentially connected. To capture this structural distinction, we extract the head and tail positions of characters within words and incorporate them into a relative positional encoding scheme. Building upon this design, we introduce Word Boundary Attention (WBA), a mechanism that assigns dynamic attention weights to characters and enhances their representations with contextual information derived from the word lattice. By explicitly modeling word boundaries, WBA effectively suppresses noise, improves word recognition, and leverages richer lexicon-based context during training. Extensive experiments across multiple datasets demonstrate that WBA consistently outperforms existing approaches, achieving, for instance, a 2.51% improvement over the base model on the Weibo dataset with YJ lexicon encoding. Furthermore, visualizations of the learned attention weights reveal the interactive relationships between words and characters, providing interpretable insights into the process of word discovery. The source code of the proposed method is publicly available at https://github.com/na978292231/WBA/tree/main/WBA4NER-main .

Resolvent analysis is used to study the low-frequency behaviour of the laminar oblique shock wave/boundary layer interaction (SWBLI). It is shown that the computed optimal gain, which can be seen as a transfer function of the system, follows a first-order low-pass filter equation, recovering the results of Touber & Sandham (J. Fluid Mech., vol. 671, 2011, pp. 417–465). This behaviour is understood as proceeding from the excitation of a single stable, steady global mode whose damping rate sets the time scale of the filter. Different Mach and Reynolds numbers are studied, covering different recirculation lengths $L$. This damping rate is found to scale as $1/L$, leading to a constant Strouhal number $St_{L}$ as observed in the literature. It is associated with a breathing motion of the recirculation bubble. This analysis furthermore supports the idea that the low-frequency dynamics of the SWBLI is a forced dynamics, in which background perturbations continuously excite the flow. The investigation is then carried out for three-dimensional perturbations for which two regimes are identified. At low wavenumbers of the order of $L$, a modal mechanism similar to that of two-dimensional perturbations is found and exhibits larger values of the optimal gain. At larger wavenumbers, of the order of the boundary layer thickness, the growth of streaks, which results from a non-modal mechanism, is detected. No interaction with the recirculation region is observed. Based on these results, the potential prevalence of three-dimensional effects in the low-frequency dynamics of the SWBLI is discussed.

In existing pre-trained language models, Chinese spelling check (CSC) often considers phonetic and graphic details at the character level, and ignores the essential role of word segmentation. To address this issue, an efficient word segmentation is studied for CSC in this paper, referred to Word Segmentation-Enhanced Speller (WOSES). The WOSES comprises two distinct models, the Word Speller (WSpeller) and the Hierarchical Word Speller (H-WSpeller), designed both for mitigating the often-ignored word boundary errors in CSC. The WOSES framework outperforms existing benchmarks on standard datasets SIGHAN13 and SIGHAN15, attributed to its innovative use of word segmentation and the specialized pre-trained model, W-MLM. Notably, the WSpeller model within the WOSES framework achieves F1 score improvements of 3.3 and 2.1% on SIGHAN13 and SIGHAN15, respectively, compared to existing methods. In this paper, the importance of word segmentation is not only underscored in CSC, but also a novel performance standard is proposed in the domain.

Text information mining is a key step to data-driven automatic/semi-automatic quality management (QM). For Chinese texts, a word segmentation algorithm is necessary for pre-processing since there are no explicit marks to define word boundaries. Because of intrinsic characteristics of QM-related texts, word segmentation algorithms for normal Chinese texts cannot be directly applied. Hence, based on the analysis of QM-related texts, we summarized six features, and proposed a hybrid Chinese word segmentation model by means of integrating transfer learning (TL), bidirectional long-short term memory (Bi-LSTM), multi-head attention (MA), and conditional random field (CRF) to construct the mTL-Bi-LSTM-MA-CRF model, considering insufficient samples of QM-related texts and excessive cutting of idioms. The mTL-Bi-LSTM-MA-CRF model is composed of two steps. Firstly, based on a word embedding space, the Bi-LSTM is introduced for context information learning, and the MA mechanism is selected to allocate attention among subspaces, and then the CRF is used to learn label sequence constraints. Secondly, a modified TL method is put forward for text feature extraction, adaptive layer weights learning, and loss function correction for selective learning. Experimental results show that the proposed model can achieve good word segmentation results with only a relatively small set of samples.

Two classic experimental paradigms – masked repetition priming and the boundary paradigm – have played a pivotal role in understanding the process of visual word recognition. Traditionally, these paradigms have been employed by different communities of researchers, with their own long-standing research traditions. Nevertheless, a review of the literature suggests that the brain-electric correlates of word processing established with both paradigms may show interesting similarities, in particular with regard to the location, timing, and direction of N1 and N250 effects. However, as of yet, no direct comparison has been undertaken between the two paradigms. In the current study, we used combined eye-tracking/EEG to perform such a within-subject comparison using the same materials (single Chinese characters) as stimuli. To facilitate direct comparisons, we used a simplified version of the boundary paradigm – the single word boundary paradigm. Our results show the typical early repetition effects of N1 and N250 for both paradigms. However, repetition effects in N250 (i.e., a reduced negativity following identical-word primes/previews as compared to different-word primes/previews) were larger with the single word boundary paradigm than with masked priming. For N1 effects, repetition effects were similar across the two paradigms, showing a larger N1 after repetitions as compared to alternations. Therefore, the results indicate that at the neural level, a briefly presented and masked foveal prime produces qualitatively similar facilitatory effects on visual word recognition as a parafoveal preview before a single saccade, although such effects appear to be stronger in the latter case.

The absence of explicit word boundaries is a distinctive characteristic of Chinese script, setting it apart from most alphabetic scripts, leading to word boundary disagreement among readers. Previous studies have examined how this feature may influence reading performance. However, further investigations are required to generate more ecologically valid and generalizable findings. In order to advance our understanding of the impact of word boundaries in Chinese reading, we introduce the Chinese Word Segmentation Agreement (CWSA) corpus. This corpus consists of 500 sentences, comprising 9813 character tokens and 1590 character types, and provides data on word segmentation agreement at each character position. The data revealed a high level of overall segmentation agreement (92%). However, participants disagreed on the position of word boundaries in 8.96% of the cases. Moreover, about 85% of the sentences contained at least one ambiguous word boundary. The character strings with high levels of disagreement were tentatively classified into three categories, namely the morphosyntactic type (e.g., “反映–了”), modifier–head type (e.g., “科學–教育”), and others (e.g., “大力–支持”). Finally, the agreement scores also significantly influenced reading behaviors, as evidenced by analyses with published eye movement data. Specifically, a high level of disagreement was associated with longer single fixation durations. We discuss the implications of these results and highlight how the CWSA corpus can facilitate future research on word segmentation in Chinese reading.

Time-resolved planar particle image velocimetry was used to analyse the structuring of a turbulent boundary layer into uniform momentum zones (UMZs). The instantaneous peak-detection method employed by Adrian et al. (J. Fluid Mech., vol. 422, 2000, pp. 1–54) and de Silva et al. (J. Fluid Mech., vol. 786, 2016, pp. 309–331) is extended to account for temporal coherence of UMZs. The resulting number of zones detected appears to follow a normal distribution at any given instant. However, the extreme cases in which the number of zones is either very high or very low, are shown to be linked with two distinct flow states. A higher than average number of zones is associated with a large-scale $Q2$ event in the log region which creates increased small-scale activity within that region. Conversely, a low number of zones corresponds to a large-scale $Q4$ event in the log region and decreased turbulent activity away from the wall. The residence times, within the measurement plane, of zones belonging to the latter scenario are shown to be on average four times larger than those of zones present during higher than average zone structuring states. For both cases, greater residence times are observed for zones of higher momentum that are generally closer to the free stream.

Improving the accuracy of handwritten character string recognition allows handwritten documents to be converted into digital text. This facilitates camera-based text input, enabling robotic process automation to manage documentation tasks. Although this field has seen significant progress, recognizing handwritten Japanese remains particularly challenging due to the difficulty of character segmentation, the wide variety of character types, and the absence of clear word boundaries. These factors make unconstrained handwritten Japanese string recognition particularly difficult for conventional approaches. Moreover, transformer-based models typically require large amounts of annotated training data. This study proposes and investigates a new String Recognition Transformer (SRT) model capable of recognizing unconstrained handwritten Japanese character strings without relying on explicit character segmentation or a large number of training images. The SRT model integrates a convolutional neural network backbone for robust local feature extraction, a Transformer encoder-decoder architecture, and a sliding window strategy that generates overlapping patches. Comparative experiments show that our method achieved a character error rate (CER) of 0.067, significantly outperforming convolutional recurrent neural network, transformer-based optical character recognition, and handwritten text recognition with Vision Transformer which achieved CERs of 0.664, 0.165, and 0.106, respectively, thereby confirming the effectiveness and robustness of the approach.