Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
16504 papers
This paper is centered around two main contributions: the first one consists in introducing several procedures for generating random dependency trees with constraints; we later use these artificial trees to compare their properties with the properties of natural trees (i.e trees extracted from treebanks) and analyze the relationships between these properties in natural and artificial settings in order to find out which relationships are formally constrained and which are linguistically motivated. We take into consideration five metrics: tree length, height, maximum arity, mean dependency distance and mean flux weight, and also look into the distribution of local configurations of nodes. This analysis is based on UD treebanks (version 2.3, Nivre et al. 2018) for four languages: Chinese, English, French and Japanese.
Cultural norms for the experience, expression, and regulation of emotion vary widely between individualistic and collectivistic cultures. Collectivistic cultures value conformity, social harmony, and social status hierarchies, which demand sensitivity and focus to broader social contexts, such that attention is directed to contextual emotion information to effectively function within constrained social roles and suppress incongruent personal emotions. By contrast, individualistic cultures valuing autonomy and personal aspirations are more likely to attend to central emotion information and to reappraise emotions to avoid negative emotional experience. Here we examined how culture affects perceptual strategies employed during emotion regulation, particularly during cognitive reappraisal and emotional suppression. Eye movements were measured while healthy young adult participants viewed negative International Affective Picture System (IAPS) images and regulated emotions by using either strategies of reappraisal (19 Asian American, 21 Caucasian American) or suppression (21 Asian American, 23 Caucasian American). After image viewing, participants rated how negative they felt as a measure of subjective emotional experience. Consistent with prior studies, reappraisers made lower negative valence ratings after regulating emotions than suppressers across both Asian American and Caucasian American groups. Although no cultural variation was observed in subjective emotional experience during emotion regulation, we found evidence of cultural variation in perceptual strategies used during emotion regulation. During middle and late time periods of emotional suppression, Asian American participants made significantly fewer fixations to emotionally salient areas than Caucasian American participants. These results indicate cultural variation in perceptual differences underlying emotional suppression, but not cognitive reappraisal.
Language standardization processes and hegemonic mechanisms are responsible for establishing the linguistic “norm” taught in schools and recognized by educated and trustworthy individuals. Through these processes and mechanisms, specific phonetic characteristics (otherwise natural to the phonetic inventory of particular languages) are identified as “non-prestigious” elements removed from the standard. Taking a sociolinguistic approach, this chapter offers a comparative review of selected phonetic characteristics, such as gheada, seseo, yeísmo, and so forth, present in Galician, Portuguese, and/or Spanish in order to shed light on the role played by institutions and/or the linguistic market in the (de-)construction of language identities.
Cross-lingual dependency parsing involves transferring syntactic knowledge\nfrom one language to another. It is a crucial component for inducing dependency\nparsers in low-resource scenarios where no training data for a language exists.\nUsing Faroese as the target language, we compare two approaches using\nannotation projection: first, projecting from multiple monolingual source\nmodels; second, projecting from a single polyglot model which is trained on the\ncombination of all source languages. Furthermore, we reproduce multi-source\nprojection (Tyers et al., 2018), in which dependency trees of multiple sources\nare combined. Finally, we apply multi-treebank modelling to the projected\ntreebanks, in addition to or alternatively to polyglot modelling on the source\nside. We find that polyglot training on the source languages produces an\noverall trend of better results on the target language but the single best\nresult for the target language is obtained by projecting from monolingual\nsource parsing models and then training multi-treebank POS tagging and parsing\nmodels on the target side.\n
Research on discourse relations between clauses, such as cause or contrast, has studied how relations are signaled in discourse. Several corpora include discourse relation annotations: the Penn Discourse Treebank (Prasad et al., 2008) annotates a subset of relations marked by explicit connectives (e.g. âhoweverâ) or understood implicit ones, while the RST-Signalling Corpus (Taboada & Das 2013) annotates the presence of signals exhaustively, but provides no information about the location of signaling devices. We present an annotation effort to anchor discourse signals at all levels, bridging the gap between these two frameworks, and support feature engineering for automatic discourse parsing.
OBJECTIVE: Cognitive-behavioral therapy (CBT) has been hypothesized to act by reducing the pathologically enhanced semantic, anxiety-related associations of patients with panic disorder. This study investigated the effects of CBT on the behavioral and neural correlates of the panic-related semantic network in patients with panic disorder. METHODS: An automatic semantic priming paradigm specifically tailored for panic disorder, in which panic symptoms (e.g., "dizziness") were primed by panic triggers (e.g., "elevator") compared with neutral words (e.g., "bottle"), was performed during functional MRI scanning with 118 patients with panic disorder (compared with 150 healthy control subjects) before and 42 patients (compared with 52 healthy control subjects) after an exposure-based CBT. Neural correlates were investigated by comparing 103 pairs of matched patients and control subjects at the baseline (for patients) or T1 (for control subjects) assessment and 39 pairs at the posttreatment or T2 assessment. RESULTS: At baseline or T1, patients rated panic-trigger/panic-symptom word pairs with higher relatedness and higher negative valence compared with healthy control subjects. Patients made faster lexical decisions to the panic-symptom words when they were preceded by panic-trigger words. This panic-priming effect in patients (compared with control subjects) was reflected in suppressed neural activation in the left and right temporal cortices and insulae and enhanced activation in the posterior and anterior cingulate cortices. After CBT, significant clinical improvements in the patient group were observed along with a reduction in relatedness and negative valence rating and attenuation of neural activation in the anterior cingulate cortex for processing of panic-trigger/panic-symptom word pairs. CONCLUSIONS: The findings support a biased semantic network in panic disorder, which is normalized after CBT. Attenuation of anterior cingulate cortex activation for processing of panic-related associations provides a potential mechanism for future therapeutic interventions.
Cross-lingual transfer is an effective way to build syntactic analysis tools in low-resource languages. However, transfer is difficult when transferring to typologically distant languages, especially when neither annotated target data nor parallel corpora are available. In this paper, we focus on methods for cross-lingual transfer to distant languages and propose to learn a generative model with a structured prior that utilizes labeled source data and unlabeled target data jointly. The parameters of source model and target model are softly shared through a regularized log likelihood objective. An invertible projection is employed to learn a new interlingual latent embedding space that compensates for imperfect crosslingual word embedding input. We evaluate our method on two syntactic tasks: part-ofspeech (POS) tagging and dependency parsing. On the Universal Dependency Treebanks, we use English as the only source corpus and transfer to a wide range of target languages. On the 10 languages in this dataset that are distant from English, our method yields an average of 5.2% absolute improvement on POS tagging and 8.3% absolute improvement on dependency parsing over a direct transfer method using state-of-the-art discriminative models. 1 3 Following Ahmad et al. ( fastText_multilingual, which contains alignment matrices for 78 languages, which also allows comparison with their numbers in Section 4.3.
Neural models have been investigated for sentiment classification over constituent trees. They learn phrase composition automatically by encoding tree structures but do not explicitly model sentiment composition, which requires to encode sentiment class labels. To this end, we investigate two formalisms with deep sentiment representations that capture sentiment subtype expressions by latent variables and Gaussian mixture vectors, respectively. Experiments on Stanford Sentiment Treebank (SST) show the effectiveness of sentiment grammar over vanilla neural encoders. Using ELMo embeddings, our method gives the best results on this benchmark.
Topic Modeling encompasses a set of techniques for text clustering and tag recommendation with significant advantages such as unsupervised learning. Based on Latent Dirichlet Allocation (LDA) topic modeling, every single word is related to a set of topics with different weight. The weights are furt her estimated in order to determine the semantic relation between the words and the rest of the documents. Apparently the chief drawback of topic modeling techniques, specifically LDA, lies on their incapability in clustering short texts in which semantic relation between words is neglected. This issue is deemed more severe when analyzing social networks such as Twitter wherein short texts are the case. It is assumed that semantic relation between a document and the target short text helps obtain efficient clustering of short texts via topic modeling. Hence, the current paper proposes a method of topic modeling named Semantic Knowledge LDA based on semantic relations between the words in tweets from Twitter social network based on the co-occurrence of words. Additionally, we propose a method of hashtag recommender system based on topic vector (TV) text similarity, named TV based Hashtag Recommender System (TVHRS). Accordingly, we applied our word co-occurrence LDA (SKLDAC) method together with WordNet lexical database to cluster the short texts from Twitter. The clustered topics are later used as the repository for the proposed hashtag recommender system. The proposed system of both SKLDA and TVHRS were applied to a set of 12,309,911 real tweets for testing purposes. When comparing the components of the proposed system to the existing methods, we recorded higher performance in terms of precision, recall and F-Score of 0.551, 0.682 and 0.526, respectively.
This paper investigates which annotation scheme of dependency treebank is more congruent for the measurement of syntactic complexity and cognitive constraint of language materials. Two representatives of semantic-and syntactic-oriented annotation schemes, the Universal Dependencies (UD) and the Surface-Syntactic Universal Dependencies (SUD), are under discussion. The results show that, on the one hand, natural languages based on both annotation schemes follow the universal linguistic law of Dependency Distance Minimization (DDM); on the other hand, according to the metric of Mean Dependency Distances (MDDs), the SUD annotation scheme that accords with traditional dependency syntaxes are more expedient to measure syntactic difficulty and cognitive demand.
The lack of understanding and definitional inconsistencies regarding agritourism and the importance of cooperation in sustaining this kind of tourism are underlined in the literature. This study analyzes the perceptions of agritourism and cooperation from actors in the sector using a plurality of methods, including unsupervised (a) text mining and (b) sentiment analysis with the use of a lexical database, as well as (c) supervised qualitative data analysis. Based on the assumption that destinations with different geographic characteristics have different features and products, two different destinations as for its accessibility and tourism recognition were selected for comparison: (a) an island—Lesvos in the North Aegean Sea, and (b) a continental mountain region—Plastiras Lake, in Greece. The data were collected from personal in-depth interviews and with the use of semi-structured questionnaires. From a methodological perspective, all three methods provided unique insights on the study’s themes, and the overall image of agritourism and cooperation was positive. A common understanding seems important for cooperation and networking; however, training is needed not only for effective promotion of agritourism, but also for cooperation techniques, benefits, trust-building mechanisms and best practices.
Recurrent Neural Networks (RNN), Long Short-Term Memory Networks (LSTM), and Memory Networks which contain memory are popularly used to learn patterns in sequential data. Sequential data has long sequences that hold relationships. RNN can handle long sequences but suffers from the vanishing and exploding gradient problems. While LSTM and other memory networks address this problem, they are not capable of handling long sequences (50 or more data points long sequence patterns). Language modelling requiring learning from longer sequences are affected by the need for more information in memory. This paper introduces Long Term Memory network (LTM), which can tackle the exploding and vanishing gradient problems and handles long sequences without forgetting. LTM is designed to scale data in the memory and gives a higher weight to the input in the sequence. LTM avoid overfitting by scaling the cell state after achieving the optimal results. The LTM is tested on Penn treebank dataset, and Text8 dataset and LTM achieves test perplexities of 83 and 82 respectively. 650 LTM cells achieved a test perplexity of 67 for Penn treebank, and 600 cells achieved a test perplexity of 77 for Text8. LTM achieves state of the art results by only using ten hidden LTM cells for both datasets.
Composers convey emotion through music by co-varying structural cues. Although the complex interplay provides a rich listening experience, this creates challenges for understanding the contributions of individual cues. Here we investigate how three specific cues (attack rate, mode, and pitch height) work together to convey emotion in Bach's Well Tempered-Clavier (WTC). In three experiments, we explore responses to (1) eight-measure excerpts and (2) musically “resolved” excerpts, and (3) investigate the role of different standard dimensional scales of emotion. In each experiment, thirty nonmusician participants rated perceived emotion along scales of valence and intensity (Experiments 1 & 2) or valence and arousal (Experiment 3) for 48 pieces in the WTC. Responses indicate listeners used attack rate, Mode, and pitch height to make judgements of valence, but only attack rate for intensity/arousal. Commonality analyses revealed mode predicted the most variance for valence ratings, followed by attack rate, with pitch height contributing minimally. In Experiment 2 mode increased in predictive power compared to Experiment 1. For Experiment 3, using “arousal” instead of “intensity” showed similar results to Experiment 1. We discuss how these results complement and extend previous findings of studies with tightly controlled stimuli, providing additional perspective on complex issues of interpersonal communication.
Many text corpora exhibit socially problematic biases, which can be propagated or amplified in the models trained on such data. For example, doctor cooccurs more frequently with male pronouns than female pronouns. In this study we (i) propose a metric to measure gender bias; (ii) measure bias in a text corpus and the text generated from a recurrent neural network language model trained on the text corpus; (iii) propose a regularization loss term for the language model that minimizes the projection of encoder-trained embeddings onto an embedding subspace that encodes gender; (iv) finally, evaluate efficacy of our proposed method on reducing gender bias. We find this regularization method to be effective in reducing gender bias up to an optimal weight assigned to the loss term, beyond which the model becomes unstable as the perplexity increases. We replicate this study on three training corpora---Penn Treebank, WikiText-2, and CNN/Daily Mail---resulting in similar conclusions.
Cross-lingual dependency parsing involves transferring syntactic knowledge from one language to another. It is a crucial component for inducing dependency parsers in low-resource scenarios where no training data for a language exists. Using Faroese as the target language, we compare two approaches using annotation projection: first, projecting from multiple monolingual source models; second, projecting from a single polyglot model which is trained on the combination of all source languages. Furthermore, we reproduce multisource projection Finally, we apply multi-treebank modelling to the projected treebanks, in addition to or alternatively to polyglot modelling on the source side. We find that polyglot training on the source languages produces an overall trend of better results on the target language but the single best result for the target language is obtained by projecting from monolingual source parsing models and then training multi-treebank POS tagging and parsing models on the target side.
Transfer parsing has been used for developing dependency parsers for languages with no treebank by using transfer from treebanks of other languages (source languages). In delexicalized transfer, parsed words are replaced by their part-of-speech tags. Transfer parsing may not work well if a language does not follow uniform syntactic structure with respect to its different constituent patterns. Earlier work has used information derived from linguistic databases to transform a source language treebank to reduce the syntactic differences between the source and the target languages. We propose a transformation method where a source language pattern is transformed stochastically to one of the multiple possible patterns followed in the target language. The transformed source language treebank can be used to train a delexicalized parser in the target language. We show that this method significantly improves the average performance of single-source delexicalized transfer parsers. We also show that, in the multi-source settings, parsers trained using a concatenation of transformed source language treebanks work better when a subset of the source language treebanks is used rather than concatenating all of them or only one. However, the problem of selecting the subset of treebanks whose combination gives the best-performing parser from the set of all the available treebanks is hard. We propose a greedy selection heuristic based on the labelled attachment scores of the corresponding single-source parsers trained using the treebanks after transformation.
Learning hierarchical abstractions from sequences is a challenging and open problem for recurrent neural networks (RNNs). This is mainly due to the difficulty of detecting features that span over long time distances with also different frequencies. In this paper, we address this challenge by introducing surprisal-based activation, a novel method to preserve activations and skip updates depending on encoding-based information content. The preserved activations can be considered as temporal shortcuts with perfect memory. We present a preliminary analysis by evaluating surprisal-based activation on language modeling with the Penn Treebank corpus and find that it can improve performance when compared to baseline RNNs and Long Short-Term Memory (LSTM) networks.
Clinical motor and non-motor effects of deep brain stimulation (DBS) of the subthalamic nucleus (STN) in Parkinson's disease (PD) seem to depend on the stimulation site within the STN. We analysed the effects of the position of the stimulation electrode within the motor STN on subjective emotional experience, expressed as emotional valence and arousal ratings to pictures representing primary rewards and aversive fearful stimuli in 20 PD patients. Patients' ratings from both aversive and erotic stimuli matched the mean ratings from a group of 20 control subjects at similar position within the STN. Patients with electrodes located more posteriorly reported both valence and arousal ratings from both the rewarding and aversive pictures as more extreme. Moreover, posterior electrode positions were associated with a higher occurrence of depression at a long-term follow-up. This brain-behavior relationship suggests a complex emotion topography in the motor part of the STN. Both valence and arousal representations overlapped and were uniformly arranged anterior-posteriorly in a gradient-like manner, suggesting a specific spatial organization needed for the coding of the motivational salience of the stimuli. This finding is relevant for our understanding of neuropsychiatric side effects in STN DBS and potentially for optimal electrode placement.
Purpose: The purpose of this paper is to prove the importance of understanding body language to achieve the effectiveness of English language classes.
 Methodology: Literature investigation is carried out to confirm the objective of this paper.
 Results: In the teaching and learning process, effective communication between a teacher and students is the utmost importance. The failure to establish effective communication in the classroom setting will result in a deficiency of the teaching and learning process.
 Implications: It is the fact that many cues of body language are culture-specific and therefore the only way to improve the understanding of body language is by interacting with people from different cultural backgrounds so that they can share socio-cultural and linguistic norms. Thus, the experience will enrich the teacher with cross-cultural nonverbal behavior which benefits his performance in the classroom. Both teachers' and students' knowledge of non-verbal language play very significant roles in making the classroom interaction successful. Therefore, finally, a summary is presented to reconfirm the importance of integrating body language into classroom interaction.
In this paper we present a novel lemmatization method based on a\nsequence-to-sequence neural network architecture and morphosyntactic context\nrepresentation. In the proposed method, our context-sensitive lemmatizer\ngenerates the lemma one character at a time based on the surface form\ncharacters and its morphosyntactic features obtained from a morphological\ntagger. We argue that a sliding window context representation suffers from\nsparseness, while in majority of cases the morphosyntactic features of a word\nbring enough information to resolve lemma ambiguities while keeping the context\nrepresentation dense and more practical for machine learning systems.\nAdditionally, we study two different data augmentation methods utilizing\nautoencoder training and morphological transducers especially beneficial for\nlow resource languages. We evaluate our lemmatizer on 52 different languages\nand 76 different treebanks, showing that our system outperforms all latest\nbaseline systems. Compared to the best overall baseline, UDPipe Future, our\nsystem outperforms it on 62 out of 76 treebanks reducing errors on average by\n19% relative. The lemmatizer together with all trained models is made available\nas a part of the Turku-neural-parsing-pipeline under the Apache 2.0 license.\n
This article presents an analysis of experiments with statistical and neural parsing techniques for Urdu, a widely spoken South Asian language. We demonstrate state of the art constituency parsing results for an Urdu treebank. Urdu is a morphologically rich and is characterized by free word order. Language representation (e.g. input type, lemmatization, word clusters), part of speech tag set, phrase labels and the size of a training corpus are crucial for parsing such languages. In this article, probabilistic context-free grammars, data-oriented parsing, and recursive neural network based models have been experimented with several linguistic features which show improvements in the parsing results. Features include syntactic sub-categorization of POS tags, empirically learned horizontal and vertical markovizations and lexical head words. These features enable dependency information for case markers and add phrasal and lexical context to the parse trees. The data-oriented parsing and recursive neural network model give an f-score of 87.1 by considering gold POS tags in the test set, on textual input, they show a performance with f-scores of 83.4 and 84.2, respectively. To overcome the issue of data sparsity due to the morphological richness, lemmatization and unsupervised word clustering have been performed. A treebank should cover most probable word orders of the language so that models can learn various orders accurately. To analyze the order coverage of the treebank and learning capability of different parsers, a test set has been prepared conditioning different word orders. This test set is evaluated with the best performing parsing models and with gold POS tags, f-scores are above 90 and on textual input, the average f-score is 87.6.
Facial expressions are fundamental to interpersonal communication, including social interaction, and allow people of different ages, cultures, and languages to quickly and reliably convey emotional information. Historically, facial expression research has followed from discrete emotion theories, which posit a limited number of distinct affective states that are represented with specific patterns of facial action. Much less work has focused on dimensional features of emotion, particularly positive and negative affect intensity. This is likely, in part, because achieving inter-rater reliability for facial action and affect intensity ratings is painstaking and labor-intensive. We use computer-vision and machine learning (CVML) to identify patterns of facial actions in 4,648 video recordings of 125 human participants, which show strong correspondences to positive and negative affect intensity ratings obtained from highly trained coders. Our results show that CVML can both (1) determine the importance of different facial actions that human coders use to derive positive and negative affective ratings when combined with interpretable machine learning methods, and (2) efficiently automate positive and negative affect intensity coding on large facial expression databases. Further, we show that CVML can be applied to individual human judges to infer which facial actions they use to generate perceptual emotion ratings from facial expressions.
Several studies have attempted to investigate how the brain codes emotional value when processing music of contrasting levels of dissonance; however, the lack of control over specific musical structural characteristics (i.e., dynamics, rhythm, melodic contour or instrumental timbre), which are known to affect perceived dissonance, rendered results difficult to interpret. To account for this, we used functional imaging with an optimized control of the musical structure to obtain a finer characterization of brain activity in response to tonal dissonance. Behavioral findings supported previous evidence for an association between increased dissonance and negative emotion. Results further demonstrated that the manipulation of tonal dissonance through systematically controlled changes in interval content elicited contrasting valence ratings but no significant effects on either arousal or potency. Neuroscientific findings showed an engagement of the left medial prefrontal cortex (mPFC) and the left rostral anterior cingulate cortex (ACC) while participants listened to dissonant compared to consonant music, converging with studies that have proposed a core role of these regions during conflict monitoring (detection and resolution), and in the appraisal of negative emotion and fear-related information. Both the left and right primary auditory cortices showed stronger functional connectivity with the ACC during the dissonant portion of the task, implying a demand for greater information integration when processing negatively valenced musical stimuli. This study demonstrated that the systematic control of musical dissonance could be applied to isolate valence from the arousal dimension, facilitating a novel access to the neural representation of negative emotion.
The affect associated with negative events fades faster than the affect associated with positive events (the fading affect bias). The fading affect bias is present in most participants and is thought to be evidence of a healthy coping mechanism operating in autobiographical memory. Prior research shows that the fading affect bias can be distorted by negative individual difference variables such as dysphoria and anxiety. The goal of this research is to link the fading affect bias to the positive individual difference variable of Grit. A total of 197 participants completed the short Grit Scale and were divided into four groups based on their Grit scores (i.e., low Grit to high Grit). Participants retrieved positive and negative event memories and then made affect ratings for the events. The results show that increased levels of Grit were associated with a stronger fading affect bias.
Recently, it has been shown that various auditory stimuli modulate flavour perception. The present study attempts to understand the effects of environmental sounds (park, food court, fast food restaurant, cafe, and bar sounds) on the perception of chocolate gelato (specifically, sweet, bitter, milky, creamy, cocoa, roasted, and vanilla notes) using the Temporal Check-All-That-Apply (TCATA) method. Additionally, affective ratings of the auditory stimuli were obtained using the Self-Assessment Manikin (SAM) in terms of their valence, arousal, and dominance. In total, 58 panellists rated the sounds and chocolate gelato in a sensory laboratory. The results revealed that bitterness, roasted, and cocoa notes were more evident when the bar, fast food, and food court sounds were played. Meanwhile, sweetness was cited more in the early mastication period when listening to park and café sounds. The park sound was significantly higher in valence, while the bar sound was significantly higher in arousal. Dominance was significantly higher for the fast food restaurant, food court, and bar sound conditions. Intriguingly, the valence evoked by the pleasant park sound was positively correlated with the sweetness of the gelato. Meanwhile, the arousal associated with bar sounds was positively correlated with bitterness, roasted, and cocoa attributes. Taken together, these results clearly demonstrate that people's perception of the flavour of gelato varied with the different real-world sounds used in this study.
Recent research on discourse relations has found that they are cued not only by discourse markers (DMs) but also by other textual signals and that signaling information is indicative of genres. While several corpora exist with discourse relation signaling information such as the Penn Discourse Treebank (PDTB, Prasad et al. 2008) and the Rhetorical Structure Theory Signalling Corpus (RST-SC, Das and Taboada 2018), they both annotate the Wall Street Journal (WSJ) section of the Penn Treebank (PTB, Marcus et al. 1993), which is limited to the news domain. Thus, this paper adapts the signal identification and anchoring scheme (Liu and Zeldes, 2019) to three more genres, examines the distribution of signaling devices across relations and genres, and provides a taxonomy of indicative signals found in this dataset.
In the past decades, linguistic typology went through a renewing phase that involved a significant change in the research questions and methods of the discipline, which is now interested in fine-grained features underlying language diversity. In this paper, we propose a novel approach to address the newly defined needs of linguistic typology by extracting qualitative and quantitative information about a wide range of features from multilingual annotated corpora based on Natural Language Processing methods and techniques. We tested our method in a case study focusing on word order variation in two widely investigated constructions, VERB-SUBJ(ect) and NOUN-ADJ(ective), with a specific view to structural and functional factors underlying the preference for one or the other order, both intra- and cross-linguistically, and their interaction. Preliminary experiments have been carried out aimed at acquiring typological evidence from a selection of linguistically annotated treebanks for three different languages, namely Italian, Spanish and English. Our results show the effectiveness of the method in letting similarities and differences also emerge from typologically close languages.
In this paper, we propose a novel approach to address the newly defined needs of linguistic typology recently interested in fine-grained features underlying language diversity. In fact, we introduce a method to extract qualitative and quantitative information about a wide range of features from multilingual annotated corpora based on Natural Language Processing methods and techniques. We tested our method in a case study focusing on word order variation in two widely investigated constructions, VERB-SUBJ(ect) and NOUN-ADJ(ective), with a specific view to structural and functional factors underlying the preference for one or the other order, both intra- and cross-linguistically, and their interaction. Preliminary experiments have been carried out aimed at acquiring typological evidence from a selection of linguistically annotated treebanks for three different languages, namely Italian, Spanish and English.
We present a novel semantic framework for modeling linguistic expressions of\ngeneralization---generic, habitual, and episodic statements---as combinations\nof simple, real-valued referential properties of predicates and their\narguments. We use this framework to construct a dataset covering the entirety\nof the Universal Dependencies English Web Treebank. We use this dataset to\nprobe the efficacy of type-level and token-level information---including\nhand-engineered features and static (GloVe) and contextual (ELMo) word\nembeddings---for predicting expressions of generalization. Data and code are\navailable at decomp.io.\n
This is the first versioned collection of.xml files containing approximately 550,000 tokens of ancient Greek prose that have been hand-analyzed into dependency syntax using Perseids/Arethusa by Prof. Vanessa Gorman of the University of Nebraska-Lincoln. CC0 1.0 license.
BACKGROUND: Neighbourhood environment characteristics have been found to be associated with residents' willingness to conduct physical activity (PA). Traditional methods to assess perceived neighbourhood environment characteristics are often subjective, costly, and time-consuming, and can be applied only on a small scale. Recent developments in deep learning algorithms and the recent availability of street view images enable researchers to assess multiple aspects of neighbourhood environment perceptions more efficiently on a large scale. This study aims to examine the relationship between each of six neighbourhood environment perceptual indicators-namely, wealthy, safe, lively, depressing, boring and beautiful-and residents' time spent on PA in Guangzhou, China. METHODS: A human-machine adversarial scoring system was developed to predict perceptions of neighbourhood environments based on Tencent Street View imagery and deep learning techniques. Image segmentation was conducted using a fully convolutional neural network (FCN-8s) and annotated ADE20k data. A human-machine adversarial scoring system was constructed based on a random forest model and image ratings by 30 volunteers. Multilevel linear regressions were used to examine the association between each of the six indicators and time spent on PA among 808 residents living in 35 neighbourhoods. RESULTS: Total PA time was positively associated with the scores for "safe" [Coef. = 1.495, SE = 0.558], "lively" [1.635, 0.789] and "beautiful" [1.009, 0.404]. It was negatively associated with the scores for "depressing" [- 1.232, 0.588] and "boring" [- 1.227, 0.603]. No significant linkage was found between total PA time and the "wealthy" score. PA was further categorised into three intensity levels. More neighbourhood perceptual indicators were associated with higher intensity PA. The scores for "safe" and "depressing" were significantly related to all three intensity levels of PA. CONCLUSIONS: People living in perceived safe, lively and beautiful neighbourhoods were more likely to engage in PA, and people living in perceived boring and depressing neighbourhoods were less likely to engage in PA. Additionally, the relationship between neighbourhood perception and PA varies across different PA intensity levels. A combination of Tencent Street View imagery and deep learning techniques provides an accurate tool to automatically assess neighbourhood environment exposure for Chinese large cities.
We present UDify, a multilingual multi-task model capable of accurately predicting universal part-of-speech, morphological features, lemmas, and dependency trees simultaneously for all 124 Universal Dependencies treebanks across 75 languages. By leveraging a multilingual BERT self-attention model pretrained on 104 languages, we found that fine-tuning it on all datasets concatenated together with simple softmax classifiers for each UD task can result in state-of-the-art UPOS, UFeats, Lemmas, UAS, and LAS scores, without requiring any recurrent or language-specific components. We evaluate UDify for multilingual learning, showing that low-resource languages benefit the most from cross-linguistic annotations. We also evaluate for zero-shot learning, with results suggesting that multilingual training provides strong UD predictions even for languages that neither UDify nor BERT have ever been trained on. Code for UDify is available at https://github.com/hyperparticle/udify.
Singlish can be interesting to the computational linguistics community both linguistically, as a major low-resource creole based on English, and computationally, for information extraction and sentiment analysis of regional social media. In our conference paper, Wang et al. (2017), we investigated part-of-speech (POS) tagging and dependency parsing for Singlish by constructing a treebank under the Universal Dependencies scheme and successfully used neural stacking models to integrate English syntactic knowledge for boosting Singlish POS tagging and dependency parsing, achieving the state-of-the-art accuracies of 89.50% and 84.47% for Singlish POS tagging and dependency, respectively. In this work, we substantially extend Wang et al. (2017) by enlarging the Singlish treebank to more than triple the size and with much more diversity in topics, as well as further exploring neural multi-task models for integrating English syntactic knowledge. Results show that the enlarged treebank has achieved significant relative error reduction of 45.8% and 15.5% on the base model, 27% and 10% on the neural multi-task model, and 21% and 15% on the neural stacking model for POS tagging and dependency parsing, respectively. Moreover, the state-of-the-art Singlish POS tagging and dependency parsing accuracies have been improved to 91.16% and 85.57%, respectively. We make our treebanks and models available for further research.
We show that sampling latent variables multiple times at a gradient step helps in improving a variational autoencoder and propose a simple and effective method to better exploit these latent variables through hidden state averaging. Consistent gains in performance on two different datasets, Penn Treebank and Yahoo, indicate the generalizability of our method. 1
Despite a great success in learning representation for image data, it is challenging to learn the stochastic latent features from natural language based on variational inference. The difficulty in stochastic sequential learning is due to the posterior collapse caused by an autoregressive decoder which is prone to be too strong to learn sufficient latent information during optimization. To compensate this weakness in learning procedure, a sophisticated latent structure is required to assure good convergence so that random features are sufficiently captured for sequential decoding. This study presents a new variational recurrent autoencoder (VRAE) for sequence reconstruction. There are two complementary encoders consisting of a long short-term memory (LSTM) and a pyramid bidirectional LSTM which are merged to discover the global and local dependencies in a hierarchical latent variable model, respectively. Experiments on Penn Treebank and Yelp 2013 demonstrate that the proposed hierarchical VRAE is able to learn the complementary representation as well as tackle the posterior collapse in stochastic sequential learning. The performance of recurrent autoencoder is substantially improved in terms of perplexity.
To develop a parser for a language with no syntactically annotated data, we either have to develop a (small) treebank for the target language or rely on cross-lingual learning or projection, or possibly use some combination of these methods. In this paper, we compare the usefulness of cross-lingual model transfer and target language annotation for three different languages, with varying support from closely related high-resource languages. The results show that annotating even a very small amount of data in the target language is superior to any cross-lingual setup and that accuracy can be further improved by adding training data from related languages in a multilingual model.
Recurrent Neural Networks (RNNs) have dominated language modeling because of their superior performance over traditional N-gram based models. In many applications, a large Recurrent Neural Network language model (RNNLM) or an ensemble of several RNNLMs is used. These models have large memory footprints and require heavy computation. In this paper, we examine the effect of applying knowledge distillation in reducing the model size for RNNLMs. In addition, we propose a trust regularization method to improve the knowledge distillation training for RNNLMs. Using knowledge distillation with trust regularization, we reduce the parameter size to a third of that of the previously published best model while maintaining the state-of-the-art perplexity result on Penn Treebank data. In a speech recognition N-best rescoring task, we reduce the RNNLM model size to 18.5% of the baseline system, with no degradation in word error rate (WER) performance on Wall Street Journal data set.
Treebank translation is a promising method for cross-lingual transfer of syntactic dependency knowledge. The basic idea is to map dependency arcs from a source treebank to its target translation according to word alignments. This method, however, can suffer from imperfect alignment between source and target words. To address this problem, we investigate syntactic transfer by code mixing, translating only confident words in a source treebank. Cross-lingual word embeddings are leveraged for transferring syntactic knowledge to the target from the resulting code-mixed treebank. Experiments on University Dependency Treebanks show that code-mixed treebanks are more effective than translated treebanks, giving highly competitive performances among cross-lingual parsing methods.
A common and mostly effective emotion regulation strategy is reappraisal. During reappraisal, activity in cognitive control brain regions increases and activity in brain regions associated with emotion responding (e.g., the amygdala) diminishes. Immediately after reappraisal, it has been observed that activity in the amygdala increases again, which might reflect a paradoxical aftereffect. While there is extensive empirical evidence for these neural correlates of emotion regulation, only few studies targeted the association with individual differences in personality traits. The aim of this study is to investigate these associations more thoroughly. Seventy-six healthy participants completed measures of broad personality traits (Big Five, Positive and Negative Affect) as well as of more narrow traits (habitual use of emotion regulation) and performed an experimental fMRI reappraisal task. Participants were instructed to either permit their emotions or to detach themselves from the presented negative and neutral pictures. After each picture, a relaxation period was included. Reappraisal success was determined by arousal ratings and activity in the amygdala. During reappraisal, we found activation in the prefrontal cortex and deactivation in the left amygdala. During the relaxation period, an immediate aftereffect was found in occipital regions and marginally in the amygdala. Neither personality traits nor habitual use of emotion regulation predicted reappraisal success or the magnitude of the aftereffect. We replicated typical activation and deactivation patterns during intentional emotion regulation and partially replicated the immediate aftereffect in the amygdala. However, there was no association between personality traits and emotion regulation success.
We propose a practical scheme to train a single multilingual sequence labeling model that yields state of the art results and is small and fast enough to run on a single CPU. Starting from a public multilingual BERT checkpoint, our final model is 6x smaller and 27x faster, and has higher accuracy than a state-of-the-art multilingual baseline. We show that our model especially outperforms on low-resource languages, and works on codemixed input text without being explicitly trained on codemixed examples. We showcase the effectiveness of our method by reporting on part-of-speech tagging and morphological prediction on 70 treebanks and 48 languages.
International audience
Seemingly, gender, language, and partisanship are intertwined concepts. We believe that the use of gendered language in political settings may be used strategically by political elites. The purpose of this paper is to craft a tool for scholars to test the interconnection between politics, gender, and language—what we refer to as being the gendered language and partisanship nexus. We test our prediction using original word rating data. From our test, we find significant variation across seven hundred words in ratings as masculine and feminine and discover that words rated as masculine are more likely to be rated as dominant and negatively valenced. We additionally find that Republican men are most likely to rate words as more masculine. Using this dictionary, we find that Republican presidents are more likely to use masculine language than Democratic presidents in their State of the Union addresses and that the Republican Party uses more masculine language than the Democratic Party in their official party platform.
Personality affect the way someone feels or acts. This paper examines the effect of personality traits, as operationalized by the Big-five questionnaire, on the number, type, and severity of the identified usability issues, physiological signals (skin conductance), and subjective emotional ratings (valence-arousal).Twenty-four users interacted with a web service and then participated in a retrospective thinking aloud session. Results revealed that the number of usability issues is significantly affected by the Openness trait. Emotional Stability significantly affects the type of reported usability issues. Problem severity is not affected by any trait. Valence ratings are significantly affected by Conscientiousness, whereas Agreeableness, Emotional Stability and Openness significantly affect arousal ratings. Finally, Openness has a significant effect on the number of detected peaks in user's skin conductance.