scholarly journals Continuous neural activations of simulation and linguistic information during metaphor processing

2021 ◽  
Author(s):  
Pei Q Liu ◽  
Louise Connell ◽  
Dermot Lynott
Keyword(s):  
Success Rate ◽  
Language Processing ◽  
Distributional Pattern ◽  
Linguistic Information ◽  
Embodied Simulation ◽  
Metaphor Comprehension ◽  
Conceptual Representations ◽  
Behavioural Study ◽  
Occurrence Pattern ◽  
Metaphor Processing

Language processing relies on conceptual representations which are composed of two crucial components, embodied simulation and linguistic distributional pattern. The embodied component refers to the reactivation of previous sensorimotor experiences related to the concept (e.g., experiences with a clever student when reading "bright student"); the linguistic component refers to the co-occurrence pattern of the constituent words (i.e., how often "bright" and "student" appear in the same context). In this study, we examined the existence and roles of these components in metaphor processing. Using both a behavioural study and EEG, we studied how these components affected the speed, success rate and neurophysiological activations of metaphor comprehension. We found that, while performance of metaphor comprehension was mainly influenced by the embodied component, the linguistic component was activated before the embodied component reached its peak and could act as a shortcut to construct good-enough representations, such that people found it easier to accept and hard to reject a metaphor when the distributional frequency of constituent words was high. In other words, the linguistic distributional pattern could provide a guide for conceptual representations before the embodied component was fully engaged.

scholarly journals An E-Prime Study on the Cognitive Mechanisms of English Predicative Metaphor Comprehension by Chinese EFL Learners

2020 ◽  
Vol 13 (10) ◽  
pp. 1
Author(s):  
Yuanlian Su ◽  
Jie Liu
Keyword(s):  
Undergraduate Students ◽  
Language Processing ◽  
Reaction Times ◽  
Cognitive Mechanisms ◽  
Embodied Simulation ◽  
Efl Learners ◽  
Priming Condition ◽  
Two Factors ◽  
Metaphor Processing ◽  
Processing Mechanism

Studies on predicative metaphors like The rumor flew through the office have not received due attention until recently. Through a behavioural experiment, this study investigates the cognitive mechanisms as well as the effects of familiarity on Chinese EFL learners’ comprehension of English predicative metaphors, adopting a two factors within-subject design: 2 (degree of familiarity: high-familiarity, low-familiarity) × 3 (priming condition: matching priming condition (MP), mismatching priming condition (MMP) and, no priming condition (NP)). Forty-five third-year Chinese undergraduate students participated in the experiment by completing a metaphor semantic comprehension test. Their reaction times (RTs) and accuracy rate of comprehension were recorded and a two-way ANOVA analysis of the results reveals that: Embodied simulation mechanism plays an important role in English predicative metaphor processing, especially when the metaphors being processed are unfamiliar. Yet its role diminishes when the metaphors being processed are highly familiar, which encourages the use of the language processing mechanism. To conclude, Chinese EFL learners make use of either the embodied simulation mechanism or the language processing mechanism in comprehending predicative metaphors, depending on their varying degrees of familiarity. These findings shed light on predicative metaphor instruction in L2 English teaching.

scholarly journals On the Role of Bilateral Brain Hypofunction and Abnormal Lateralization of Cortical Information Flow as Neural Underpinnings of Conventional Metaphor Processing Impairment in Schizophrenia: An fMRI and EEG Study

2021 ◽  
Author(s):  
Przemysław Adamczyk ◽  
Martin Jáni ◽  
Tomasz S. Ligeza ◽  
Olga Płonka ◽  
Piotr Błądziński ◽  
...  
Keyword(s):  
Information Flow ◽  
Language Processing ◽  
Time Course ◽  
Figurative Language ◽  
Neural Circuits ◽  
Effective Connectivity ◽  
Brain Regions ◽  
Blood Oxygenation ◽  
Control Group ◽  
Metaphor Processing

AbstractFigurative language processing (e.g. metaphors) is commonly impaired in schizophrenia. In the present study, we investigated the neural activity and propagation of information within neural circuits related to the figurative speech, as a neural substrate of impaired conventional metaphor processing in schizophrenia. The study included 30 schizophrenia outpatients and 30 healthy controls, all of whom were assessed with a functional Magnetic Resonance Imaging (fMRI) and electroencephalography (EEG) punchline-based metaphor comprehension task including literal (neutral), figurative (metaphorical) and nonsense (absurd) endings. The blood oxygenation level-dependent signal was recorded with 3T MRI scanner and direction and strength of cortical information flow in the time course of task processing was estimated with a 64-channel EEG input for directed transfer function. The presented results revealed that the behavioral manifestation of impaired figurative language in schizophrenia is related to the hypofunction in the bilateral fronto-temporo-parietal brain regions (fMRI) and various differences in effective connectivity in the fronto-temporo-parietal circuit (EEG). Schizophrenia outpatients showed an abnormal pattern of connectivity during metaphor processing which was related to bilateral (but more pronounced at the left hemisphere) hypoactivation of the brain. Moreover, we found reversed lateralization patterns, i.e. a rightward-shifted pattern during metaphor processing in schizophrenia compared to the control group. In conclusion, the presented findings revealed that the impairment of the conventional metaphor processing in schizophrenia is related to the bilateral brain hypofunction, which supports the evidence on reversed lateralization of the language neural network and the existence of compensatory recruitment of alternative neural circuits in schizophrenia.

FindThatQuote: A Question-Answering Web-based System to Locate Quotes using Deep Learning and Natural-Language Processing

2021 ◽  
Author(s):  
Nathan Ji ◽  
Yu Sun
Keyword(s):  
Deep Learning ◽  
Natural Language Processing ◽  
Natural Language ◽  
Success Rate ◽  
Language Processing ◽  
Question Answering ◽  
Specific Information ◽  
Web Based ◽  
User Friendly ◽  
Web Based System

The digital age gives us access to a multitude of both information and mediums in which we can interpret information. A majority of the time, many people find interpreting such information difficult as the medium may not be as user friendly as possible. This project has examined the inquiry of how one can identify specific information in a given text based on a question. This inquiry is intended to streamline one's ability to determine the relevance of a given text relative to his objective. The project has an overall 80% success rate given 10 articles with three questions asked per article. This success rate indicates that this project is likely applicable to those who are asking for content level questions within an article.

scholarly journals Detection of Gene Interactions Based on Syntactic Relations

2008 ◽  
Vol 2008 ◽  
pp. 1-9 ◽  
Author(s):  
Mi-Young Kim
Keyword(s):  
Language Processing ◽  
Training Dataset ◽  
Phase Method ◽  
Gene Interactions ◽  
Second Phase ◽  
Good Precision ◽  
Linguistic Information ◽  
Three Phase ◽  
Syntactic Relations ◽  
Learning Language

Interactions between proteins and genes are considered essential in the description of biomolecular phenomena, and networks of interactions are applied in a system's biology approach. Recently, many studies have sought to extract information from biomolecular text using natural language processing technology. Previous studies have asserted that linguistic information is useful for improving the detection of gene interactions. In particular, syntactic relations among linguistic information are good for detecting gene interactions. However, previous systems give a reasonably good precision but poor recall. To improve recall without sacrificing precision, this paper proposes a three-phase method for detecting gene interactions based on syntactic relations. In the first phase, we retrieve syntactic encapsulation categories for each candidate agent and target. In the second phase, we construct a verb list that indicates the nature of the interaction between pairs of genes. In the last phase, we determine direction rules to detect which of two genes is the agent or target. Even without biomolecular knowledge, our method performs reasonably well using a small training dataset. While the first phase contributes to improve recall, the second and third phases contribute to improve precision. In the experimental results using ICML 05 Workshop on Learning Language in Logic (LLL05) data, our proposed method gave an F-measure of 67.2% for the test data, significantly outperforming previous methods. We also describe the contribution of each phase to the performance.

Developmental Steps in Metaphorical Language Abilities: The Influence of Age, Gender, Cognitive Flexibility, Information Processing Speed, and Analogical Reasoning

2017 ◽  
Vol 62 (2) ◽  
pp. 207-228 ◽  
Author(s):  
Ulrike Willinger ◽  
Matthias Deckert ◽  
Michaela Schmöger ◽  
Ines Schaunig-Busch ◽  
Anton K Formann ◽  
...  
Keyword(s):  
Information Processing ◽  
Cognitive Flexibility ◽  
Early Adolescence ◽  
Processing Speed ◽  
Analogical Reasoning ◽  
Time Pressure ◽  
Early Adolescents ◽  
Information Processing Speed ◽  
Metaphor Comprehension ◽  
Metaphor Processing

Purpose: Metaphor is a specific type of figurative language that is used in various important fields such as in the work with children in clinical or teaching contexts. The aim of the study was to investigate the developmental course, developmental steps, and possible cognitive predictors regarding metaphor processing in childhood and early adolescence. Method: One hundred sixty-four typically developing children (7-year-olds, 9-year-olds) and early adolescents (11-year-olds) were tested for metaphor identification, comprehension, comprehension quality, and preference by the Metaphoric Triads Task as well as for analogical reasoning, information processing speed, cognitive flexibility under time pressure, and cognitive flexibility without time pressure. Results: Metaphor identification and comprehension consecutively increased with age. Eleven-year-olds showed significantly higher metaphor comprehension quality and preference scores than seven- and nine-year-olds, whilst these younger age groups did not differ. Age, cognitive flexibility under time pressure, information processing speed, analogical reasoning, and cognitive flexibility without time pressure significantly predicted metaphor comprehension. Conclusions: Metaphorical language ability shows an ongoing development and seemingly changes qualitatively at the beginning of early adolescence. These results can possibly be explained by a greater synaptic reorganization in early adolescents. Furthermore, cognitive flexibility under time pressure and information processing speed possibly facilitate the ability to adapt metaphor processing strategies in a flexible, quick, and appropriate way.

scholarly journals The Neural Basis of Metaphor Comprehension: Evidence from Left Hemisphere Degeneration

2020 ◽  
Vol 1 (4) ◽  
pp. 474-491
Author(s):  
Nathaniel Klooster ◽  
Marguerite McQuire ◽  
Murray Grossman ◽  
Corey McMillan ◽  
Anjan Chatterjee ◽  
...  
Keyword(s):  
Neurodegenerative Disease ◽  
Left Hemisphere ◽  
Figurative Language ◽  
Inferior Frontal Gyrus ◽  
Neural Systems ◽  
Neural Basis ◽  
Language Abilities ◽  
Metaphor Comprehension ◽  
Imaging Research ◽  
Metaphor Processing

Despite the ubiquity of metaphor in cognition and communication, it is absent from standard clinical assessments of language, and the neural systems that support metaphor processing are debated. Previous research shows that patients with focal brain lesions can display selective impairments in processing metaphor, suggesting that figurative language abilities may be disproportionately vulnerable to brain injury. We hypothesized that metaphor processing is especially vulnerable to neurodegenerative disease, and that the left hemisphere is critical for normal metaphor processing. To evaluate these hypotheses, we tested metaphor comprehension in patients with left-hemisphere neurodegeneration, and in demographically matched healthy comparison participants. Stimuli consisted of moderately familiar metaphors and closely matched literal sentences sharing the same source term (e.g., The interview was a painful crawl / The infant’s motion was a crawl). Written sentences were presented, followed by four modifier-noun answer choices (one target and three foils). Healthy controls, though reliably better at literal than metaphor trials, comprehended both sentence conditions well. By contrast, participants with left-hemisphere neurodegeneration performed disproportionately poorly on metaphor comprehension. Anatomical analyses show relationships between metaphor accuracy and patient atrophy in the left middle and superior temporal gyri, and the left inferior frontal gyrus, areas that have been implicated in supporting metaphor comprehension in previous imaging research. The behavioral results also suggest deficits of metaphor comprehension may be a sensitive measure of cognitive dysfunction in some forms of neurodegenerative disease.

scholarly journals On the Linguistic Representational Power of Neural Machine Translation Models

2020 ◽  
Vol 46 (1) ◽  
pp. 1-52
Author(s):  
Yonatan Belinkov ◽  
Nadir Durrani ◽  
Fahim Dalvi ◽  
Hassan Sajjad ◽  
James Glass
Keyword(s):  
Machine Translation ◽  
Language Processing ◽  
Lexical Semantics ◽  
Linguistic Information ◽  
Neural Machine Translation ◽  
Recent Success ◽  
Part Of Speech ◽  
Morphologically Rich Languages ◽  
Representational Power ◽  
Semantic Dependencies

Despite the recent success of deep neural networks in natural language processing and other spheres of artificial intelligence, their interpretability remains a challenge. We analyze the representations learned by neural machine translation (NMT) models at various levels of granularity and evaluate their quality through relevant extrinsic properties. In particular, we seek answers to the following questions: (i) How accurately is word structure captured within the learned representations, which is an important aspect in translating morphologically rich languages? (ii) Do the representations capture long-range dependencies, and effectively handle syntactically divergent languages? (iii) Do the representations capture lexical semantics? We conduct a thorough investigation along several parameters: (i) Which layers in the architecture capture each of these linguistic phenomena; (ii) How does the choice of translation unit (word, character, or subword unit) impact the linguistic properties captured by the underlying representations? (iii) Do the encoder and decoder learn differently and independently? (iv) Do the representations learned by multilingual NMT models capture the same amount of linguistic information as their bilingual counterparts? Our data-driven, quantitative evaluation illuminates important aspects in NMT models and their ability to capture various linguistic phenomena. We show that deep NMT models trained in an end-to-end fashion, without being provided any direct supervision during the training process, learn a non-trivial amount of linguistic information. Notable findings include the following observations: (i) Word morphology and part-of-speech information are captured at the lower layers of the model; (ii) In contrast, lexical semantics or non-local syntactic and semantic dependencies are better represented at the higher layers of the model; (iii) Representations learned using characters are more informed about word-morphology compared to those learned using subword units; and (iv) Representations learned by multilingual models are richer compared to bilingual models.

scholarly journals The neural basis of metaphor comprehension: evidence from left hemisphere degeneration

2020 ◽  
Author(s):  
Nathaniel Klooster ◽  
Margeurite McQuire ◽  
Murray Grossman ◽  
Corey McMillan ◽  
Anjan Chatterjee ◽  
...  
Keyword(s):  
Neurodegenerative Disease ◽  
Left Hemisphere ◽  
Figurative Language ◽  
Inferior Frontal Gyrus ◽  
Neural Basis ◽  
Language Abilities ◽  
Metaphor Comprehension ◽  
Imaging Research ◽  
Healthy Control ◽  
Metaphor Processing

Despite the ubiquity of metaphor in cognition and communication, it is absent from standard clinical assessments of language, and the neural systems that support metaphor processing are debated. Previous research shows that patients with focal brain lesions can display selective impairments in processing metaphor, suggesting that figurative language abilities may be disproportionately vulnerable to brain injury. We hypothesized that metaphor processing is especially vulnerable to neurodegenerative disease, and that the left hemisphere is critical for normal metaphor processing. To evaluate these hypotheses, we tested metaphor comprehension in patients with left-hemisphere neurodegeneration (LHND), and in demographically- matched healthy control subjects (HC). Stimuli consisted of moderately-familiar metaphors and closely matched literal sentences sharing the same source term (The interview was a painful crawl/ The infant’s motion was a crawl). Written sentences were presented, followed by four modifier-noun answer choices (one target and three foils). HC, though reliably better at literal than metaphor trials, comprehended both sentence conditions well. By contrast, LHND participants performed disproportionately poorly on metaphor comprehension. Anatomical analyses show relationships between metaphor accuracy and patient atrophy in left middle and superior temporal gyri, and the left inferior frontal gyrus, areas that have been implicated in supporting metaphor comprehension in previous imaging research. The behavioral results also suggest deficits of metaphor comprehension may be a sensitive measure of cognitive dysfunction in some forms of neurodegenerative disease.

scholarly journals Textual Backdoor Defense via Poisoned Sample Recognition

2021 ◽  
Vol 11 (21) ◽  
pp. 9938
Author(s):  
Kun Shao ◽  
Yu Zhang ◽  
Junan Yang ◽  
Hui Liu
Keyword(s):  
Success Rate ◽  
Language Processing ◽  
Training Data ◽  
Infection Model ◽  
Search Range ◽  
Word Level ◽  
Sentence Level ◽  
Preliminary Model ◽  
Sample Recognition ◽  
Better Than

Deep learning models are vulnerable to backdoor attacks. The success rate of textual backdoor attacks based on data poisoning in existing research is as high as 100%. In order to enhance the natural language processing model’s defense against backdoor attacks, we propose a textual backdoor defense method via poisoned sample recognition. Our method consists of two parts: the first step is to add a controlled noise layer after the model embedding layer, and to train a preliminary model with incomplete or no backdoor embedding, which reduces the effectiveness of poisoned samples. Then, we use the model to initially identify the poisoned samples in the training set so as to narrow the search range of the poisoned samples. The second step uses all the training data to train an infection model embedded in the backdoor, which is used to reclassify the samples selected in the first step, and finally identify the poisoned samples. Through detailed experiments, we have proved that our defense method can effectively defend against a variety of backdoor attacks (character-level, word-level and sentence-level backdoor attacks), and the experimental effect is better than the baseline method. For the BERT model trained by the IMDB dataset, this method can even reduce the success rate of word-level backdoor attacks to 0%.

scholarly journals Argot: Generating Adversarial Readable Chinese Texts

2020 ◽  
Author(s):  
Zihan Zhang ◽  
Mingxuan Liu ◽  
Chao Zhang ◽  
Yiming Zhang ◽  
Zhou Li ◽  
...  
Keyword(s):  
Image Processing ◽  
Natural Language Processing ◽  
Success Rate ◽  
Language Processing ◽  
Black Box ◽  
Essential Step ◽  
Chinese Characteristics ◽  
Adversarial Examples ◽  
Chinese Texts ◽  
Chinese And English

Natural language processing (NLP) models are known vulnerable to adversarial examples, similar to image processing models. Studying adversarial texts is an essential step to improve the robustness of NLP models. However, existing studies mainly focus on analyzing English texts and generating adversarial examples for English texts. There is no work studying the possibility and effect of the transformation to another language, e.g, Chinese. In this paper, we analyze the differences between Chinese and English, and explore the methodology to transform the existing English adversarial generation method to Chinese. We propose a novel black-box adversarial Chinese texts generation solution Argot, by utilizing the method for adversarial English samples and several novel methods developed on Chinese characteristics. Argot could effectively and efficiently generate adversarial Chinese texts with good readability. Furthermore, Argot could also automatically generate targeted Chinese adversarial text, achieving a high success rate and ensuring readability of the Chinese.

Sign in / Sign up

Export Citation Format

Share Document