Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
16504 papers
Abstract Kankana-ey is a widely used dialect in the northern region of the Philippines. Unfortunately, there are documented studies on the syntactic rules of this dialect. This study explored the development of a corpus for the Kankana-ey dialect. Further, the corpus was then used to establish the syntactic rules of Kankana-ey. A Kankana-ey version of the bible, dictionaries, news articles, songs and various online resources were used to collect words for the corpus of the Kankana-ey dialect. These identified words were also tagged using the parts of speech tags of the Penn TreeBank. Using the corpus and TensorFlow, 320 Kankana-ey sentences were analysed to determine the syntactic rules. In addition, 80 sentences were used to test the accuracy of the identified rules. At the end of the study, the created corpus has 3,412 tagged Kankana-ey words, while the analysis of the syntactic rules resulted to 1,722 rules. Testing also showed a 60% accuracy of the syntactic rules. In conclusion, the high number of identified rules from the 320 sentences was due to multiple Kankana-ey words having different possible tags. This also resulted to the low accuracy of the syntactic rules.
This article discusses one of the forms of machine translation, the Instagram translation feature called “see translation”. The research is focused on the translation techniques applied by the machine in translating Banyumas batik motifs from Indonesian to English found in @batikantodjamil and @batk_rd. This topic is worth discussing since machine translation is now getting more developed and is projected to replace human translator. However, in some cases, for example in dealing with culturally-bound terms, machine translation cannot perform contextual knowledge as well as the human translator. this mini research was conducted by applying qualitative research with purposive sampling technique in which the researchers obtain the data by selecting two batik center Instagram accounts containing batik motif names in the captions. The result shows that there are three translation techniques applied by the Instagram translation features, namely literal, borrowing, and particularization. The most dominant technique to use is borrowing technique, and it shows a tendency that such cultural terms in the source language do not have one-to-one correspondence in the target language. In other words, the touch of human translator is very important in the post-editing process of translation by machine to make the translation more acceptable. However, if it is impossible to involve human translator, the Instagram administrator should enrich the machine with more contextual linguistic database to provide the users with better translation results.
The Western Canadian Dictionary and Phrase-Book (WCD) was written in 1912 as a guide for British immigrants who were encountering a variety of English that was “more resistant to British linguistic norms than the conservative Anglophone heartland of Ontario” (Considine 2003: 252). Though the dictionary has been researched in terms of its lexicographical value, relatively little research has examined the historical and cultural reasons as to why a dictionary of western Canadian English was viable at the time it was written. This thesis examines the connections between the WCD and the Canadian Government’s pre-World War I immigration campaign. This includes connections between the writer of the dictionary, John Sandilands, and the Canadian Government through Sandilands position as a proofreader of pamphlets intended to advertise the West that were produced by the Department of the Interior. The thesis also examines how the dictionary participates in a network of literature produced at the time to reproduce a “Promised Land” (Francis 1989) narrative of the Canadian West which became “the dominant perception of the region during the formative years of agricultural settlement” (Francis & Kitzan 2007: IX). This network of literature includes stories by Nellie McClung and poetry by Robert J. C. Stead, who both employ Promised Land narratives in their work. Within these narratives a western Canadian dialect, marked by slang found in the WCD, becomes associated with a heightened morality of its speakers, and is used as a shorthand for values of hard work and honesty. Ultimately, it is argued that the dictionary reflects a dominant, settler, narrative of the West that was pushed by the Canadian Government to ‘sell’ the West.
Abstract Consumerism is an inherent feature of a modern consumer-minded society which enhances in some people both a hunger for collecting and a more serious desire developing into oniomania (shopping mania, shopaholism), kleptomania or pathological hoarding (syllogomania, Diogenes syndrome, Plyushkin’s syndrome). The paper proposes an interdisciplinary approach to the problem of pathological hoarding of unnecessary things and domestic animals by tenants of condominiums in Russian cities. Socio-legal prerequisites for this psychosocial disease still insufficiently studied in the country are also analyzed in the paper. The data of the Federal State Statistics Service of the Russian Federation and various legal acts were examined. In the course of the study, methods of quality content analysis and visual sociology were used to analyze cases of pathological hoarding highlighted in Russian digital media in recent years. The Clutter-Hoarding Scale and Clutter Image Rating Scale were used to interpret photos of cluttered Russian flats in condos. In conclusion, recommendations are given on improving state policy and the legislation of the Russian Federation.
First study in series of three tracking attention during encoding and subsequent memory accuracy after presenting misinformation. Study one using static images. Project covers eye tracking data, memory accuracy scores, mood data, valence and arousal ratings, study documentation and stimuli, assessment of central and peripheral information in visual medium.
Response Time Data and Valence Rating Data. <br>Covariate Tests: Big 5 Test, D2-R, MWT-B<br>Group 0 = Zen group, Group 1 = comparison group
Recurrent neural networks (RNNs) are a widely used deep architecture for\nsequence modeling, generation, and prediction. Despite success in applications\nsuch as machine translation and voice recognition, these stateful models have\nseveral critical shortcomings. Specifically, RNNs generalize poorly over very\nlong sequences, which limits their applicability to many important temporal\nprocessing and time series forecasting problems. For example, RNNs struggle in\nrecognizing complex context free languages (CFLs), never reaching 100% accuracy\non training. One way to address these shortcomings is to couple an RNN with an\nexternal, differentiable memory structure, such as a stack. However,\ndifferentiable memories in prior work have neither been extensively studied on\nCFLs nor tested on sequences longer than those seen in training. The few\nefforts that have studied them have shown that continuous differentiable memory\nstructures yield poor generalization for complex CFLs, making the RNN less\ninterpretable. In this paper, we improve the memory-augmented RNN with\nimportant architectural and state updating mechanisms that ensure that the\nmodel learns to properly balance the use of its latent states with external\nmemory. Our improved RNN models exhibit better generalization performance and\nare able to classify long strings generated by complex hierarchical context\nfree grammars (CFGs). We evaluate our models on CGGs, including the Dyck\nlanguages, as well as on the Penn Treebank language modelling task, and achieve\nstable, robust performance across these benchmarks. Furthermore, we show that\nonly our memory-augmented networks are capable of retaining memory for a longer\nduration up to strings of length 160.\n
How do each of us come to view the world uniquely? An emerging theory of microvalence proposes that subtle feelings of reward and punishment derived from individualized experiences with basic everyday objects help determine how we later attend and behave towards them. These objects that are part of our more mundane experiences are thought to be given attentional priority similar to objects that evoke stronger emotional responses. However, this relationship between preferences guided by daily experience and attention has not been tested. I introduced a novel paradigm to induce microvalences by simulating real life experience paired with an interocular suppression technique (bCFS) to explore its role in attention. Consistent with the theory of microvalence, affective ratings indicated that our novel shapes possessed pre-existing affective properties by which they are evaluated, giving rise to preferences. Unexpectedly, we observed a unifying effect of experience, blurring perceived differences between novel shapes, thus collapsing initial preferences (feelings of like or dislike). Results showed, however, that microvalences were not prioritized in attention. Our findings place emphasis on the role of experience in shifting automatic preferences to create unbiased representations of the world.
Acceleration and wide deployability in deeper recurrent neural network is hindered by high demand for computation and memory storage on devices with memory and latency constraints. In this work, we propose a novel regularization method to learn hardware-friendly sparse structures for deep recurrent neural networks. Considering the consistency of dimension in continuous time units in recurrent neural networks, low-rank structured sparse approximations of the weight matrices are learned through the regularization without dimension distortion. Our method is architecture agnostic and can learn compact models with higher degree of sparsity than the state-of-the-art structured sparsity learning method. The structured sparsity rather than random sparsity also facilitates the hardware implementation. Experiments on language modeling of Penn TreeBank dataset show that our approach can reduce the parameters of stacked recurrent neural network model by over 90% with less than 1% perplexity loss. It is also successfully evaluated on larger highway neural network model with word2vec dataset like enwik8 and text8 using only 20M weights.
A growing number of studies have shown that, compared to young adults, older adults better remember positive information than negative information. However, it is not clear whether this age-related positivity effect relies on an increase in positive information memory and/or on a decrease in negative information memory. Thus, we aimed to study the specific mechanisms underlying the age-related positivity effect in different memory tasks. To do so, we used an emotional word memory paradigm including immediate free recall, recognition and delayed free recall tasks. Forty-five young adults (m = 20.0 years) and 45 older adults (m = 69.2 years) native French speakers participated. Thirty-six low French words, including 12 negative (e.g. égout), 12 positive (e.g. lagune) and 12 neutral (e.g. notion) words were selected from an emotional lexical database (Gobin et al. 2017). For the recognition task, 36 new words were selected. The results showed that the age-related positivity effect specifically depended on a decrease in negativity preference (i.e. the comparison between negative and neutral words) in older adults, in comparison with young adults, both in immediate and delayed free recall tasks. Indeed, in these tasks, young adults recalled more negative than neutral words whereas there was no difference in older adults. In recognition task, no age-related positivity effect has been observed. Moreover, the results showed that, in immediate recall, the higher the older adults memory abilities, the lower their negativity preference. This correlation was not significant in delayed recall. These results suggest that, when compared with young adults, older adults disengage from negative words processing through costly cognitive processes. A small magnitude of negativity preference would indicate good maintenance of memory abilities. Results are discussed in the framework of the socioemotional selectivity theory.
Literary works are classified as works of imagination in the form of fictional or imaginary experiences. The messages to be conveyed through literary works must be creative so that they appear attractive to read and listen to, so it is necessary to have a stile from the authorship itself to make his work beautiful and attractive. There are many ways to enjoy, understand and appreciate the work of the author Tenas Effendy, one of which is by studying the Stile of Tenas Effendy's authorship in Tunjuk Ajar Melayu. This study aims to analyze and interpret Tenas Effendy's stile authorship in Tunjuk Ajar Melayu. This needs to be examined because the existence of a literary work can be seen from how the author packs his work so that he can create his own stile from the author's side. The method used in this research is content analysis method. The source of data in this research is Tunjuk Ajar Melayu Karya Tenas Effendy in 2013 which has been recorded. The data collection technique is done by applying the hermeneutic technique. After the author of the analysis, Tunjuk Ajar Melayu by Tenas Effendy has a unique and distinctive authorship stile seen from the stylistic aspect, namely the stile as a pack of thoughts, the stile as a deviation from linguistic norms and the stile as a collection of personal characteristics.
The article highlights monitoring and dynamics of language development in post-soviet Kyrgyzstan. The crucial politico-social changes after the collapse of the USSR affected greatly the language situation of the country. Hence, this article’s aim is to give some sociolinguistic analysis to the following issues: • To monitor country’s up-to date language situation • To speak on the new language policy of Kyrgyzstan in modern stage • To encounter new socio-economic and political challenges in the process of language lawmaking • To define the interrelationships between dominant languages and vernaculars of minority communities This article also explains the reasons of granting Russian language the status of Lingua Franca, immediately after Kyrgyzstan’s becoming an independent state. One of the acute problems today is corpus planning, which means codification of newly coined or borrowed words. The flow of new terminology from the other languages, especially “Americanisms” and “Englishisisms” which replaced so-called “Sovietisms”, needs to be standardized according to the linguistic norms of state (Kyrgyz) language. The article also reveals “hierarchical” disposition of main languages due to their functional load in 20 domains of Kyrgyzstan. For the last 28 years’ tremendous changes have happened in language space of Kyrgyzstan, directly touching upon the positions of Kyrgyz, Russian, Uzbek, English, Turkish and other languages. Therefore, language system of Kyrgyzstan nowadays presents complex, intertwined interrelations between all nationalities residing in Kyrgyzstan and language cooperation among them. It happened because of some socio-political reasons: 1) The flow of Russian speaking population from industrial areas to the Russian Federation. 2) Changing demographic situation on the country. 3) Inner immigration process when many unemployed people from the distant regions came to Bishkek and Chui valley to find job possibilities.
We analyse and explain the increased generalisation performance of iterate averaging using a Gaussian process perturbation model between the true and batch risk surface on the high dimensional quadratic. We derive three phenomena \latestEdits{from our theoretical results:} (1) The importance of combining iterate averaging (IA) with large learning rates and regularisation for improved regularisation. (2) Justification for less frequent averaging. (3) That we expect adaptive gradient methods to work equally well, or better, with iterate averaging than their non-adaptive counterparts. Inspired by these results\latestEdits{, together with} empirical investigations of the importance of appropriate regularisation for the solution diversity of the iterates, we propose two adaptive algorithms with iterate averaging. These give significantly better results compared to stochastic gradient descent (SGD), require less tuning and do not require early stopping or validation set monitoring. We showcase the efficacy of our approach on the CIFAR-10/100, ImageNet and Penn Treebank datasets on a variety of modern and classical network architectures.
Cultural priming studies frequently employ non-validated, stereotypical images. Here, we select images to separately evoke two cultural mindsets: Hispanic and US-American. Spanish-English bilinguals identifying as Hispanic/Latino (N=149) rated 50 images online for their cultural and emotional evocation. Based on relative cultural identification, cultural "delegate" (strongly US-American, strongly Hispanic, balanced bicultural) subsamples' ratings were averaged to isolate particularly salient images. Image ratings were compared across respondents' national origins. Ratings of seven selected pairs of content-matched Hispanic and US-American primes were compared across the full sample. High discrimination across cultural mindsets and positive emotion ratings were maintained regardless of various demographic factors. Thus, we provide empirical justification for incorporating these stimuli, individually or as sets, within cultural priming studies among Hispanic/Latino samples.
Semantic Role Labelling (SRL) is the process of automatically finding the semantic roles of terms in a sentence. It is an essential task towards creating a machine-meaningful representation of textual information. One public linguistic resource commonly used for this task is the FrameNet Project. FrameNet is a human and machine-readable lexical database containing a considerable number of annotated sentences, those annotations link sentence fragments to semantic frames. However, while the annotations across all the documents covered in the dataset link to most of the frames, a large group of frames lack annotations in the documents pointing to them. In this paper, we present a data augmentation method for FrameNet documents that increases by over 13% the total number of annotations. Our approach relies on lexical, syntactic, and semantic aspects of the sentences to provide additional annotations. We evaluate the proposed augmentation method by comparing the performance of a state-of-the-art semantic-role-labelling system, trained using a dataset with and without augmentation.
'Synonym' is an imperative instrument of commonsense knowledge that we apply to make a good sense and sound judgement of our reading. To investigate the ability of machine comprehension models in handling the synonym commonsense knowledge, we developed an innovative approach to automatically generate a dataset based on the Stanford Question Answering Dataset (SQuAD 2.0). The brand-new dataset consists of additional distracting sentences or questions spawned using synonym commonsense knowledge. We formulated new questions by replacing noun entities of the original ones in SQuAD 2.0 with their synonyms. This approach followed the two fundamental principles of SQuAD 2.0 dataset: relevancy and plausibility (incorrect answers are more challenging if they are relevant and plausible). It improves the robustness/abstraction of the question set. To improve the synonym selection strategy in Word Sense Disambiguation (WSD) problem, we designed a new algorithm Multiple Source Adapted Lesk Algorithm (MSALA). Rather than only using WordNet as the source of gloss for adapted Lesk algorithm, we used both lexical database WordNet and commonsense database ConceptNet. This fusion provides a rich hierarchy of semantic relations for the MSALA algorithm. Using this method, we devised 11,000 questions and evaluated the performance of the state-of-the-art question answering system-BERT. Our result shows that the accuracy of the contemporary BERT-Base model dropped from 74.98% to 63.24%. This 10+% accuracy drop revealed the limitations of BERT in handling synonym commonsense knowledge.
Text discourse parsing plays an important role in understanding information flow and argumentative structure in natural language. Previous research under the Rhetorical Structure Theory (RST) has mostly focused on inducing and evaluating models from the English treebank. However, the parsing tasks for other languages such as German, Dutch, and Portuguese are still challenging due to the shortage of annotated data. In this work, we investigate two approaches to establish a neural, cross-lingual discourse parser via: (1) utilizing multilingual vector representations; and (2) adopting segment-level translation of the source content. Experiment results show that both methods are effective even with limited training data, and achieve state-of-the-art performance on cross-lingual, document-level discourse parsing on all sub-tasks.
This paper investigates whether typical stress patterns in English nouns and verbs are available as a prosodic cue for categorisation and accelerated word learning during first language acquisition. The stress typicality hypothesis states that left-stressed nouns and right-stressed verbs should be acquired earlier than the reverse configurations if stress effectively signals lexical class membership. In this view, class-typical stress patterns are expected to facilitate learning of novel items. A series of generalized additive models (GAMs) based on a comprehensive set of lexical data (CELEX) as well as a large set of age-of-acquisition (AoA) and concreteness ratings reveals that stress typicality plays a minor role in early acquisition, as it is generally superseded by a preference for left-hand (or 'trochaic') patterns in both nouns and verbs. This may be explained by general cognitive constraints (perceptual salience and recency) or exposure to the dominant pattern in the ambient language.
An interesting and frequent type of multiword expression (MWE) is the headless MWE, for which there are no true internal syntactic dominance relations; examples include many named entities ("Wells Fargo") and dates ("July 5, 2020") as well as certain productive constructions ("blow for blow", "day after day").Despite their special status and prevalence, current dependency-annotation schemes require treating such flat structures as if they had internal syntactic heads, and most current parsers handle them in the same fashion as headed constructions.Meanwhile, outside the context of parsing, taggers are typically used for identifying MWEs, but taggers might benefit from structural information.We empirically compare these two common strategies-parsing and tagging-for predicting flat MWEs.Additionally, we propose an efficient joint decoding algorithm that combines scores from both strategies.Experimental results on the MWE-Aware English Dependency Corpus and on six non-English dependency treebanks with frequent flat structures show that: (1) tagging is more accurate than parsing for identifying flat-structure MWEs, (2) our joint decoder reconciles the two different views and, for non-BERT features, leads to higher accuracies, and (3) most of the gains result from feature sharing between the parsers and taggers.
Deep learning has promoted remarkable progress in various tasks while the effort devoted to these hand-crafting neural networks has motivated so-called neural architecture search (NAS) to discover them automatically. Recent aging evolution (AE) automatic search algorithm turns to discard the oldest model in population and finds image classifiers beyond manual design. However, it achieves a low speed of convergence. A nonaging evolution (NAE) algorithm tends to neglect the worst architecture in population to accelerate the search process whereas it obtains a lower performance compared with AE. To address this issue, in this letter, we propose to use an optimized evolution algorithm for recurrent NAS (EvoRNAS) by setting a probability ϵ to remove the worst or oldest model in population alternatively, which can balance the performance and time length. Besides, parameter sharing mechanism is introduced in our approach due to the heavy cost of evaluating the candidate models in both AE and NAE. Furthermore, we train the sharing parameters only once instead of many epochs like ENAS, which makes the evaluation of candidate models faster. On Penn Treebank, we first explore different ϵ in EvoRNAS and find the best value suited for the learning task, which is also better than AE and NAE. Second, the optimal cell found by EvoRNAS can achieve state-of-the-art performance within only 0.6 GPU hours, which is 20 × and 40 × faster than ENAS and DARTS. Moreover, the transferability of the learned architecture to WikiText-2 also shows strong performance compared with ENAS or DARTS.
Sentiment analysis, especially for long documents, plausibly requires methods\ncapturing complex linguistics structures. To accommodate this, we propose a\nnovel framework to exploit task-related discourse for the task of sentiment\nanalysis. More specifically, we are combining the large-scale,\nsentiment-dependent MEGA-DT treebank with a novel neural architecture for\nsentiment prediction, based on a hybrid TreeLSTM hierarchical attention model.\nExperiments show that our framework using sentiment-related discourse\naugmentations for sentiment prediction enhances the overall performance for\nlong documents, even beyond previous approaches using well-established\ndiscourse parsers trained on human annotated data. We show that a simple\nensemble approach can further enhance performance by selectively using\ndiscourse, depending on the document length.\n
In order to extract the semantic and grammatical information of sentences more effectively, this paper proposes a sentence sentiment classification method based on Self-supervised and Self-attention mechanism (SS-SAtt-BiLSTM). In this method, BiLSTM network is used to extract the feature of text context relationship, and self-supervised (SS) learning mode is introduced into the supervised sentence representation model. The sentence itself is used as the label data information of current words, and an improved self-attention mechanism (SA) is used to calculate the attention weight of each moment. The experimental results of MR and Stanford sentient treebank (sst-5) data sets show that this method reduces the dependence on tagged data, and the improved self-attention mechanism enables the model to learn more key features of sentences and improve the classification performance.
In neural machine translation (NMT), sequence distillation (SD) through creation of distilled corpora leads to efficient (compact and fast) models.However, its effectiveness in extremely low-resource (ELR) settings has not been well-studied.On the other hand, transfer learning (TL) by leveraging larger helping corpora greatly improves translation quality in general.This paper investigates a combination of SD and TL for training efficient NMT models for ELR settings, where we utilize TL with helping corpora twice: once for distilling the ELR corpora and then during compact model training.We experimented with two ELR settings: Vietnamese-English and Hindi-English from the Asian Language Treebank dataset with 18k training sentence pairs.Using the compact models with 40% smaller parameters trained on the distilled ELR corpora, greedy search achieved 3.6 BLEU points improvement in average while reducing 40% of decoding time.We also confirmed that using both the distilled ELR and helping corpora in the second round of TL further improves translation quality.Our work highlights the importance of stage-wise application of SD and TL for efficient NMT modeling for ELR settings.
The position of the word final –s, after a weakening in archaic Latin, seems to be fixed in the spoken language in the classical period. Then, it partially disappeared in the Romance languages: in modern languages, it is conserved only north and west of the Massa–Senigallia line, while we cannot find it neither in the eastern regions nor in South Italy. Based on this fact, linguists generally claim that the weakening of the final –s started only after the intensive dialectal diversification of Latin, simultaneously with the evolution of the Romance languages. However, the data of the Computerized Historical Linguistic Database of Latin Inscriptions of the Imperial Age (LLDB) do not verify this generally accepted opinion. We can find almost as many examples of the lack of word final –s as that of –m also from the earlier centuries of the Imperial age. The aim of this paper is to explore the reasons behind the inconsistencies between the scholarly consensus and the epigraphical data.
We present a method for conducting morphological disambiguation for South Sámi, which is an endangered language. Our method uses an FST-based morphological analyzer to produce an ambiguous set of morphological readings for each word in a sentence. These readings are disambiguated with a Bi-RNN model trained on the related North Sámi UD Treebank and some synthetically generated South Sámi data. The disambiguation is done on the level of morphological tags ignoring word forms and lemmas; this makes it possible to use North Sámi training data for South Sámi without the need for a bilingual dictionary or aligned word embeddings. Our approach requires only minimal resources for South Sámi, which makes it usable and applicable in the contexts of any other endangered language as well.
RST-based discourse parsing is an important NLP task with numerous downstream applications, such as summarization, machine translation and opinion mining. In this paper, we demonstrate a simple, yet highly accurate discourse parser, incorporating recent contextual language models. Our parser establishes the new state-of-the-art (SOTA) performance for predicting structure and nuclearity on two key RST datasets, RST-DT and Instr-DT. We further demonstrate that pretraining our parser on the recently available large-scale "silver-standard" discourse treebank MEGA-DT provides even larger performance benefits, suggesting a novel and promising research direction in the field of discourse analysis.
The article analyzes differences in the description of discourse relations in corpus research, in particular with the reference to the use of discourse markers – expressions that tie together subsequent fragments of the text and provide information about the nature of these relations. The text presents three concepts of the description of explicitness and implicitness of the content: Rhetorical Structure Theory, Penn Discourse Treebank and the author’s original proposal and indicates the consequences of each solution. The analysis of relations with particles as metatexual expressions defined in accordance with The Nest Dictionary of Polish reveals the possibility of expressing explicitness as a representation of elements of informational structure shaped by the use of a given particle, and implicitness as a lack of representation of certain elements of this type.
Current models of dialogue mainly focus on utterances within a topically coherent discourse segment, rather than new-topic utterances (NTUs), which begin a new topic not correlating with the content of prior discourse. As a result, these models may sufficiently account for discourse context of task-oriented but not social conversations. We conduct a pilot annotation study of NTUs as a first step towards a model capable of rationalizing conversational coherence in social talk. We start with the naturally occurring social dialogues in the Disco-SPICE corpus, annotated with discourse relations in the Penn Discourse Treebank (PDTB) and Cognitive approach to Coherence Relations (CCR) frameworks. We first annotate content-based coherence relations that are not available in Disco-SPICE, and then heuristically identify NTUs, which lack a coherence relation to prior discourse. Based on the interaction between NTUs and their discourse context, we construct a classification for NTUs that actually convey certain non-topical coherence in social talk. This classification introduces new sequence-based social intents that traditional taxonomies of speech acts do not capture. The new findings advocates the development of a Bayesian game-theoretic model for social talk. 1
Named Entity Recognition is an information extraction technique that identifies name entities in a text. Three popular methods have been conventionally used namely: rule-based, machine-learning-based and hybrid of them to extract named entities from a text. Machine-learning-based methods have good performance in the Persian language if they are trained with good features. To get good performance in Conditional Random Field-based Persian Named Entity Recognition, a several syntactic features based on dependency grammar along with some morphological and language-independent features have been designed in order to extract suitable features for the learning phase. In this implementation, designed features have been applied to Conditional Random Field to build our model. To evaluate our system, the Persian syntactic dependency Treebank with about 30,000 sentences, prepared in NOOR Islamic science computer research center, has been implemented. This Treebank has Named-Entity tags, such as Person, Organization and location. The result of this study showed that our approach achieved 86.86% precision, 80.29% recall and 83.44% F-measure which are relatively higher than those values reported for other Persian NER methods.
Noun phrases convey key information in communication and are of interest in NLP tasks. A base NP is defined as the headword and left-hand side modifiers of a noun phrase. In this thesis, we identify base NPs in Universal Dependencies treebanks in English and French using an RNN architecture.The data of this thesis consist of three multi-layered treebanks in which each sentence is annotated in both constituency and dependency formalisms. To build our training data, we find base NPs in the constituency layers and project them onto the dependency layer by labeling corresponding tokens. For input features, we devised 18 configurations of features available in UD annotation. We train RNN models with LSTM and GRU cells with different numbers of epochs on these configurations of features.Tested on monolingual and bilingual test sets, our models delivered satisfactory token-based F1 scores (92.70% on English, 94.87% on French, 94.29% on bilingual test set). The most predicative configuration of features is found out to be pos_dep_parent_child_morph, which covers 1) dependency relations between the current token, its syntactic head, its leftmost and rightmost syntactic dependents; 2) PoS tags of these tokens; and 3) morphological features of the current token.
As the number, size, and complexity of building construction projects increase, code compliance checking becomes more challenging because of the time-consuming, costly, and error-prone nature of a manual checking process. A fully automated code compliance checking would be desirable in facilitating a more efficient, cost effective, and human error-proof code checking. Such automation requires automated information extraction from building designs and building codes, and automated information transformation to a format that allows automated reasoning. Natural language processing (NLP) is an important technology to support such automated processing of building codes, because building codes are represented in natural language texts. Part-of-speech (POS) tagging, as an important basis of NLP tasks, must have a high performance to ensure the quality of the automated processing of building codes in such a compliance checking system. However, no systematic testing of existing POS taggers on domain specific building codes data have been performed. To address this gap, the authors analyzed the performance of seven state-of-the-at POS taggers on tagging building codes and compared their results to a manually-labeled gold standard. The authors aim to: (1) find the best performing tagger in terms of accuracy, and (2) identify common sources of errors. In providing the POS tags, the authors used the Penn Treebank tagset, which is a widely used tagset with a proper balance between conciseness and information richness. An average accuracy of 88.80% was found on the testing data. The Standford coreNLP tagger outperformed the other taggers in the experiment. Common sources of errors were identified to be: (1) word ambiguity, (2) rare words, and (3) unique meaning of common English words in the construction context. The found result of machine taggers on building codes calls for performance improvement, such as error-fixing transformational rules and machine taggers that are trained on building codes.
Neural machine translation (NMT) models are typically trained using a softmax cross-entropy loss where the softmax distribution is compared against smoothed gold labels. In low-resource scenarios, NMT models tend to over-fit because the softmax distribution quickly approaches the gold label distribution. To address this issue, we propose to divide the logits by a temperature coefficient, prior to applying softmax, during training. In our experiments on 11 language pairs in the Asian Language Treebank dataset and the WMT 2019 English-to-German translation task, we observed significant improvements in translation quality by up to 3.9 BLEU points. Furthermore, softmax tempering makes the greedy search to be as good as beam search decoding in terms of translation quality, enabling 1.5 to 3.5 times speed-up. We also study the impact of softmax tempering on multilingual NMT and recurrently stacked NMT, both of which aim to reduce the NMT model size by parameter sharing thereby verifying the utility of temperature in developing compact NMT models. Finally, an analysis of softmax entropies and gradients reveal the impact of our method on the internal behavior of NMT models.
The Neural Machine-Parsed IcePaHC is a machine-parsed treebank which consists of Icelandic texts from the 13th to 20th century, mostly Icelandic sagas. The texts were parsed using the IceNeuralParsingPipeline, a parsing pipeline which includes an Icelandic model of the Berkeley Neural Parser along with pre- and postprocessing steps. The parser was trained on IcePaHC and the parsing scheme of the treebank is therefore the same, although the treebank does not include empty phrases or lemmas. The treebank includes 52 texts. The total word count is 1,716,429 and the total number of clauses is 167,815.
New language phenomena are driven by social and political shifts at the global level. Even though the traditional literary norm is being destroyed, these linguistic innovations fulfil a language compensatory function. Internet communication and the new speech processes found in it provoke a keen research interest and are extensively explored by linguists. Major global changes in our life (cloud-based technologies, ecology, post-truth, the problem of generations, Big Data, etc.) were bound to transform communication itself. Therefore, we see changes in genres, functional styles, texts and our traditional ideas of various forms of the Russian national language usage. The Russian Internet (Runet) reveals language potential, fulfils the compensatory function of the language filling in all the elements missing so far and language shortcomings (neologisms denoting feminine gender-specific job titles, deviant verbal forms, new structures in comparative forms of adverbs and adjectives, etc.). The speech system of the Internet communication should be considered not as a double-sided one (oral and written) but as a conceptually new digital form of language use. In the democratic environment of pluralism, tolerance and the freedom of language use, lexical and lexical-grammatical innovations, “the new vernacular”, irregular grammar and lexical collocability, as well as the direct and conscious intention to break the norm of the literary language, should be justified and deemed a manifestation of the compensatory language function. Special attention is given to the acute problem of fundamental transformations in teaching practice.
The article is a study of phonetic-morphological and semantic-stylistic features of argotic vocabulary in contemporary French language. The work is devoted to questions of structural-semantic and phonetic-morphological peculiarities of the argotic vocabulary in printed mass media. The conducted research has shown that the argotic vocabulary can be considered as a special lexical subsystem of spoken language, which is characterized by a steady tendency to penetrate into higher linguistic levels due to its phonetic-morphological and semantic features. The systematization of phonetic and semantic processes in the argotic dictionary is carried out. We have noticed that argot from a linguistic point of view is an expressive lexical subsystem of spoken language, which is characterized by a large expressive potential and rapid changes in vocabulary and penetrates into higher levels of the French language. From the point of view of modern linguistics, the French language can be viewed in vertical and horizontal sections. Horizontal division is caused by the existence of the dialectal partition of the French language. The vertical division is explained by the existence of social groups that use one or another sociolect. Analyzing the evolution of the definition of argot, we can assume that in its development argot passed the long way from the language taboo to the special lexical subsystem of the literary-spoken language. We have noticed that it is important to distinguish argot and spoken language. In spite of the both scientific and practical interest in the spoken language problem and the emergence of numerous studies that led to the creation of colloquialism as a special section of linguistics, a number of aspects of spoken language (approaches to its identification, differentiation of spoken language and related phenomena) remain insufficiently highlighted. The spoken language (vernacular), according to modern linguistic assertions, occupies an intermediate position between the spoken-literary language, dialects and sociolects. We have noticed that the democratization of the norms of the literary language led to the emergence of a literary-colloquial form of spoken language. Such a combination of literary and spoken language is caused by the nature of social development. Key words: argot; slang; French language; dialect; argotheistic vocabulary; non-normative variant elements; semantics; phonetics; morphology; mass media.
В данной статье проведен этнолингвистический анализ 53 наименований ягодных растений в якутском языке с целью определения их способов образования. Языковой материал был взят из различных лексикографических источников, в том числе ботанических словарей. Эмпирической базой исследования послужили также полевые и экспедиционные наблюдения, диалектные записи и материалы, собранные авторами с 2017 г. в различных районах Республики Саха (Якутия). С целью определения лексического значения корней анализируемых наименований приведены семантические описания лексем из «Большого толкового словаря якутского языка», «Словаря якутского языка» Э. К. Пекарского. Для выявления способов образования фитонимов использованы методы словообразовательного анализа: поиск производящей основы, выделение словообразовательного форманта, установление принципов словообразования. Также применены элементы семантического и морфологического анализа наименований, рассмотрены синтаксические конструкции сложных слов (словосочетаний). Непроизводным лексическим единицам даны этимологические характеристики, в некоторых случаях приведены параллели из других языков. Заимствованные названия ягодных растений проанализированы с точки зрения фонетических изменений, такой же принцип анализа использован относительно диалектных единиц, перешедших в литературную норму. Кроме того, некоторые якутские основы и рефлексы в других родственных языках сравнены в фоноструктурном аспекте. Путем описательного метода даны биологические характеристики растений, также приведены объяснения географическим особенностям мест произрастания исследуемых ягодных растений. Установлено, что основным способом номинации фитонимов, обозначающих наименования ягодных растений в якутском языке, является номинация по признаку, в основе которой лежит ряд мотивационных характеристик. Удалось выяснить, что данный принцип основывается на ассоциативной метафоризации, которая раскрывает отношение якутов к живой природе и отражает особенность мировоззрения народа саха. Наличие заимствований из русского языка объясняется общей территорией проживания якутов и русских. Фитонимы монгольского и тунгусо-маньчжурского происхождения свидетельствует о тесных языковых контактах якутов с монгольскими племенами и тунгусо-маньчжурскими народами, в частности эвенками. In this article, an ethnolinguistic analysis of 53 names of berry plants in the Yakut language was carried out to determine their methods of formation. Language material was taken by their various lexicographic sources, including botanical dictionaries. The empirical basis of the study was also field and expeditionary observations, dialect records and materials collected by the authors since 2017 in various regions of the Republic of Sakha (Yakutia). To determine the lexical value of the roots of the analyzed names, semantic descriptions of lexemes are given from the «Great Thick Dictionary of the Yakut Language» and the «Dictionary of the Yakut Language» by E. K. Pekarskiy. Methods of word-formation analysis are used to identify methods of phytonym formation: search of producing base, extraction of word-formation formant, the establishment of word formation principles. Elements of semantic and morphological analysis of nominations are also used, syntax constructions of complex words (phrases) are considered. Non-derivative lexical units are given etymological characteristics, in some cases, parallels from other languages are given. Borrowed names of berry plants are analyzed in terms of phonetic changes, the same principle of analysis is used relative to dialect units that have converted to the literary norm. Besides, some Yakut bases and reflexes in other related languages are compared in the phonostructural aspect. Biological characteristics of plants are given through descriptive method, as well as explanations of geographical peculiarities of places of growth of studied berry plants are given. It has been established that the main method of nomination of phytonyms, which denote names of berry plants in the Yakut language, his nomination based on several motivational characteristics. It was possible to find out that this principle is based on associative metaphorization, which reveals the attitude of Yakuts to wildlife and reflects the characteristics of the world view of the Sakha people. The availability of borrowing from the Russian language is explained by the general territory of residence of Yakuts and Russians. Phytonyms of Mongolian and Tunguso-Manchurian origins indicate close linguistic contacts of Yakuts with Mongolian tribes and Tunguso-Manchurian peoples, particularly Evens.
The article deals with the issues of the application of the theory of genres while handling professional medical discourses. The practical implementation of the theory of genres in learning a professional medical discourse is one of the efficient techniques in current methodology of teaching English. The topicality of the research is determined, firstly, by the fact that different genres are commonly used in the practice of medical professional interaction and the domain of a foreign language teaching in a medical language classroom. Secondly, there is a need to outline grammar and lexical peculiarities of the genres of medical discourses and their functioning in oral and written speech. The authors study the problems of the theory of genres concerning medical discourses. The researchers characterize the main lexical and grammar peculiarities of the genre of an oral presentation of a patient’s medical history, the genre of a scientific text for the specialty “Technologies of Medical Diagnosing and Treatment”, the genre of a letter of referral to a medical specialist and reveal specific features of their functioning in oral and written speech. Linguistic analysis of a medical discourse which is focused on lexical, grammatical, functional and rhetoric peculiarities of a written or an oral discourse appears to be very fruitful allowing us to answer the questions how tests are structured and organized. Textual success often depends on the familiarity of text organization for readers of the discourse community, however small or large that community might be and students should be aware of the genre norms. Dealing with professional medical genres, we assume that every genre is clearly structured and has its own sets of language structures. The task of language instructors is to acquaint the learners with existing genre norms. The authors have singled out the “moves” of the three genres and developed the sequence of pre-reading, while-reading, post-reading and information gap activities which are effective while working on the discourses that have been studied.
The article is based on an understanding of the possible functional and semantic classification of lexical units of the Portuguese language according to a graded criterion, which corresponds to the operational perspective of the semantic application of the language. One of the forms of visualization of grading operations can be a grading scale, or a graduation scale, the possibility of which is supported by the existence of an intuitive perception of a certain sample, a certain point of reference, a certain norm, above and below which are certain zones of units that fall into the grading situation. The author notes that the grading operator as a minimal linguistic variable is not only a marker that specifies the degree of deviation from a certain ordinary level and provides a modification of the value (movement down or up the axiological scale), but also an element of ordering reasoning, expression of opinion, and personal attitude of the Portuguese speaker. The article analyzes operators that belong to the group of high-degree and ultimate-measure graduators. The analysis of the combinability of the operators considered by the author allowed us to distinguish two ways of grading limit features in the Portuguese language: ingerent and extensive. Extensive gain has more to do with the verb, in the amplification of which the orientation of the actants are expressed more explicitly. This allows you to select a special type of gain – actant gain. However, even when grading adjectives, some Portuguese ultimate-measure gradators or graduators are able to participate in extensive models, such as the quantifier pronoun todo, toda (all, entire, whole). In addition to differences in the method of modifying a trait (extensive or inherent) and in the modal part of the value, ultimate measure operators differ in the nature of the trait representation. Some of them represent a trait in statics, regardless of its previous development (absolutamente, inteiramente, totalmente), and others represent the ultimate measure of the trait as the result of its previous development and accumulation (completamente, todo, de todo).
Summary This paper compares the romanization of Gaul in the 1st century BC and the gallicization of the island of Martinique during 17th-century French colonial expansion, using criteria set out by Muf- wene's Founder Principle. The Founder Principle determines key ecological factors in the formation of creole vernaculars, such as the founding populations and their proportion to the whole, language varieties spoken, and the nature and evolution of the interactions of the founding populations (also referred to as “colonization styles”). Based on the comparison, it will be claimed that new languages arise when a language undergoes vehicularization and subsequently shifts from one speech community to another. In other words, linguistic genesis would be a complicated case of language contact, where not only one, but sev- eral dialects of both superstrate and substrate varieties are involved, in a historical context where the identity function of language, or the norm, is overriden by the need to communicate. Research also indicates that language varieties spoken at the time of the shift did not pertain to normative usage, but to popular varieties, dialects, or both, since the emerging vernaculars - in Gaul, as well as in Martinique - preserved some of their phonological and lexical particularities.
The article presents a study of the consequences of digitalization of modern Russian education and public life in terms of its impact on the Russian language and its native speakers. In the course of it two opposite estimates were revealed. One point of view considers this influence destructive, making the language primitive, not independent and filled with borrowings. The other one considers the Internet and digitalization as a whole as a means of developing, enriching and updating the Russian language, making it live and modern. Purism and anti-purism are also evident in the differences between conflict and non-conflict-related practices of digital hygiene in the field of communication, including those in educational environments. The Internet language and Internet discourse are either excluded from educational environments or are becoming one of the trends of their development in the conditions of digitalization. The fundamental principle of the research is the principle of anthropocentrism (as opposed to media centrism), which is expressed in the idea of language as a practice that implements the connection between a person and the world, where the media act only as intermediaries, tools. The leading method is trendwatching, i.e. detecting trends in the transformation of the Russian language through content analysis of the blogosphere -a special communication space on the Internet. In addition, the paper used a structural approach and transformational analysis to the study of language, which allowed us to identify the features of the blogosphere, where the Internet discourse unfolds. Among the features of the blogosphere the following ones were noted: cognitivity, interactivity, variability, polyphony, creativity, simulation and hyperreality. The descriptive method was used to determine the mechanisms of changing the Russian language under the influence of the Internet, including: lexical, represented by neologisms and erratives; semantic, including abbreviations and acronyms; and cognitive, which include metaphors, metonymies and lituratives. The selected mechanisms, on the one hand, expand the boundaries of cognitive existence, and on the other, lead to the loss of the possibility of full-fledged transmission and understanding of meanings. Finally, the pragmalinguistic method allowed us to justify the need to develop constructive practices of digital hygiene in the field of network and non-network communication, to determine their repertoire, characteristics and educational potential. The author concludes that digital hygiene practices based on the idea of the conventionality of network and non-network communication norms are insufficiently studied and widespread in educational environments, and that one-dimensional ideas about the vandal influence of Internet discourse on the Russian language are rejected.
Abstract This article is a synthesis of the major elements of a sociolinguistic theory presented by Jean Le Dû and Yves Le Berre in their recent book, Métamorphoses, Trente ans de sociolinguistique à Brest (1984–2014). Given that both authors come from native Breton-speaking families in Western Brittany and have experienced the language shift to French first-hand, they provide a unique, inside view of the process as well as the reasons Breton speakers opted in favour of French. The sociolinguistic concepts they have imagined provide highly useful tools that highlight the inseparable bond between language and the social, political and economic forces that govern our choices. More specifically, they point out that the “Breton language” is splintered into as many varieties as there are social and geographic entities in western Brittany. For this reason, it should not be viewed as a monolithic entity. Far from “reviving” or “saving” the language, the authors argue that the recent creation of a phonologically, grammatically and lexically unified Breton norm is often so distant from the vernacular language that it has provoked a new form of diglossia which failed to reverse the break in the transmission of the natural language. The book provides tremendous insight into the complex issues which lead people to shift to another language. Language planners and scholars working on similar endangered language situations and who want to understand the mechanisms at work (and thus hopefully have some success in their endeavours) would do well to take heed of their experience.
This paper presents a multilingual legal information retrieval system for mapping recitals to articles in European Union (EU) directives and normative provisions in national legislation. Such a system could be useful for purposive interpretation of norms. A previous work on mapping recitals and normative provisions was limited to EU legislation in English and only one lexical text similarity technique. In this paper, we develop state-of-the-art text similarity models to investigate the interplay between directive recitals, directive (sub-)articles and provisions of national implementing measures (NIMs) on a multilingual corpus (from Ireland, Italy and Luxembourg). Our results indicate that directive recitals do not have a direct influence on NIM provisions, but they sometimes contain additional information that is not present in the transposed directive sub-article, and can therefore facilitate purposive interpretation.
The key problems of modern linguistics are the establishment and description of the relationship between the semantic foundations of the language, the national mentality, and the culture of a particular ethnic group. At this point, ethnospecific vocabulary is of great interest to linguists, as well as for representatives of cultural studies, sociology, and ethnography. In this regard, this article is aimed to identify the features of national currency nominations that have a number of quantitative and qualitative characteristics in the German language, which is manifested at the level of frequency of use, as well as in terms of combinability at the level of syntagmatics. The methodological basis of the study were the main provisions of the theory of cognitive linguistics and the use of corpus technologies in the analysis of lexical units. The author compared data from paremiographic sources with data from the German language corpus "Digitales Wrterbuch der deutschen Sprache" (DWDS) on ethnospecific lexical units with subsequent identification of cognitive-discursive features of the analyzed group of lexemes. The obtained data can be taken into account when describing the mechanisms of formation and maintenance of national identity, translation of norms and values in the German-speaking ethnocultural society.
Generic masculines – masculine forms used for women – are employed in many languages, for example English (Mills 2008), French (Coady 2018), Spanish (Bengoechea 2011) and German (Motschenbacher 2016), providing accounts of how gender is made visible in the language through morphological, lexical and syntactic units. These accounts are also linked with how gender is seen in societies and culture, reproducing an imbalance between women and men. Specifically, language discrimination against women is based on the idea that speakers orient themselves towards androcentric language, recognising ‘men’ as a metonym for the group ‘human being’ (Alvanoudi 2014), causing a linguistic invisibility of women. Similarly, studies in Italian have also discussed the use of masculine forms to refer to, talk about and describe women (Cavagnoli 2013), or have shown how these are used in specialised (Nardone 2016, 2018) or media corpora (Formato 2014, 2016, 2019). This article investigates the use of a specific (and underexamined) generic masculine in Italian – namely, the indefinite pronoun uno.m.sg (in comparison with una.f.sg) labelled ‘impersonal masculine’ (Formato 2019:69) – in three subcorpora of the Perugia Corpus (TV, Web and Spoken; Spina 2014). Uno.m.sg is seen as constructing ‘extended intersubjectivity’, that is, the awareness of a general third party (3rdP) acting as the social bearer of the utterance (Tantucci 2013, 2016, 2017a). The results show that the masculine impersonal uno.m.sg is widely used in the three subcorpora and in several functions, confirming that grammatically gendered language is still employed within a ‘masculine as a norm’ order.
This article is devoted to the development of a linguistic model for describing the concept WOMAN. The material is women’s dialect discourse. The sources of the material are the Tomsk dialect corpus which includes materials of expeditions organized by dialectologists of Tomsk State University from 1946 to the present days on the territory of Middle Ob dialects spread. In the article we used modeling method based on the idea of the nominative field of a concept, as well as an interpretation technique relying on analysis of contexts, and a method of quantitative calculations used in relation to units that represent the concept. Lexical and phraseological units that make up the nominative field of the concept were revealed during the research. These units were divided into the following lexical-semantic groups: 1) the general nominations of a female person; 2) age and status in marriage; 4) status in the family hierarchy; 5) anatomical and biological characteristics; 6) character traits and behavior; 7) appearance characteristics; 8) profession and work processes. Elements of different layers of the concept are revealed in each lexical- semantic group. All of them give a general picture of ideas about women. So, the basis for identifying of gender conceptualizations and stereotypes is the presence of linguistic oppositions of male and female; the presence of a large number of lexical units that reflect the status of marriage (girl, bride, young woman, wife, mistress, old woman, widow, old girl, brooch and so on); lexical pairs that are opposed to each other on the basis of evaluation “positive” – “negative” (clean, clean – dirty, mistress – disheveled, etc.). A large number of words that negatively assess certain qualities and behavior of women (gossip girl, market woman, stramovka, etc.) indicate the high requirements imposed on the woman, the condemnation of deviations from social norms. The content of the concept of WOMAN depends on the specifics of rural existence, which is based on work, the presence of patriarchal gender stereotypes, social and historical events and processes. The significance of the research is determined by the possibility of using its results for development of a new interdisciplinary scientific field – gender dialectology that studies the gender characteristics of the dialect.
The purpose of this study was to find out the correlation between the level of academic background and the rules of Hangul orthography by examining the compliance of Korean spelling. To examine this, six universities were divided into three divisions by level into A, B, and C, and the status of marking by university was compared with the free bulletin board of Everytime, a college student community. As a result of the survey, A-grade K universities best observed the Hangul Hangul orthography followed by C-grade A universities. The university that did not keep the Hangul orthography well was a C-grade D university. By educational level, the grade with the lowest mislabeling rate is grade A. It can be seen that the status of compliance with the lexical norms is related to each level of education. However, it cannot be generalized because there are not many vocabulary in common from six universities, but there is some correlation between knowledge and actual notation.
This study aims to examine how humour can be used as a communication strategy in a crisis communication work with an objective of creating crisis awareness among the target audience and through this, contribute to the research field of Strategic communication and digital media. Research concerning humour as a strategy combined with risk communication is yet limited and therefore this paper has the ambition to contribute with new knowledge about whether humour as a strategy is appropriate and successful when transmitting a preparatory crisis message, that can be seen as a topic difficult to relate with for the target audience. The empirical material is limited to a digital advertising campaign. The campaign was launched in December 2019 by the Swedish Civil Contingencies Agency (MSB) on Swedish television and social media channels and consists of three videos from the campaign. Based on theories that concerns national risk- and crisis communication, humour as a strategy, national humour and social norms, a multimodal critical discourse analysis (MCDA) has been implemented on the empirical material to find out whether the producer’s lexical choices, representation of the characters and power relations can contribute with knowledge that regards if humour can work as a strategy, in a situation where a crisis doesn’t exists yet. The result shows that humour can work as a strategy if it is being used with caution and if the producer takes the specific context, culture and the target audience's earlier experiences of crisis into consideration, when adapting the preparatory message.